diff options
Diffstat (limited to 'src/VBox/Debugger/DBGPlugInLinux.cpp')
| -rw-r--r-- | src/VBox/Debugger/DBGPlugInLinux.cpp | 3045 |
1 files changed, 3045 insertions, 0 deletions
diff --git a/src/VBox/Debugger/DBGPlugInLinux.cpp b/src/VBox/Debugger/DBGPlugInLinux.cpp new file mode 100644 index 00000000..89ae306b --- /dev/null +++ b/src/VBox/Debugger/DBGPlugInLinux.cpp @@ -0,0 +1,3045 @@ +/* $Id: DBGPlugInLinux.cpp $ */ +/** @file + * DBGPlugInLinux - Debugger and Guest OS Digger Plugin For Linux. + */ + +/* + * Copyright (C) 2008-2023 Oracle and/or its affiliates. + * + * This file is part of VirtualBox base platform packages, as + * available from https://www.virtualbox.org. + * + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public License + * as published by the Free Software Foundation, in version 3 of the + * License. + * + * This program is distributed in the hope that it will be useful, but + * WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, see <https://www.gnu.org/licenses>. + * + * SPDX-License-Identifier: GPL-3.0-only + */ + + +/********************************************************************************************************************************* +* Header Files * +*********************************************************************************************************************************/ +#define LOG_GROUP LOG_GROUP_DBGF /// @todo add new log group. +#include "DBGPlugIns.h" +#include "DBGPlugInCommonELF.h" +#include <VBox/vmm/vmmr3vtable.h> +#include <VBox/dis.h> +#include <iprt/ctype.h> +#include <iprt/file.h> +#include <iprt/err.h> +#include <iprt/mem.h> +#include <iprt/stream.h> +#include <iprt/string.h> +#include <iprt/vfs.h> +#include <iprt/zip.h> + + +/********************************************************************************************************************************* +* Structures and Typedefs * +*********************************************************************************************************************************/ + +/** @name InternalLinux structures + * @{ */ + + +/** @} */ + + +/** + * Config item type. + */ +typedef enum DBGDIGGERLINUXCFGITEMTYPE +{ + /** Invalid type. */ + DBGDIGGERLINUXCFGITEMTYPE_INVALID = 0, + /** String. */ + DBGDIGGERLINUXCFGITEMTYPE_STRING, + /** Number. */ + DBGDIGGERLINUXCFGITEMTYPE_NUMBER, + /** Flag whether this feature is included in the + * kernel or as a module. */ + DBGDIGGERLINUXCFGITEMTYPE_FLAG +} DBGDIGGERLINUXCFGITEMTYPE; + +/** + * Item in the config database. + */ +typedef struct DBGDIGGERLINUXCFGITEM +{ + /** String space core. */ + RTSTRSPACECORE Core; + /** Config item type. */ + DBGDIGGERLINUXCFGITEMTYPE enmType; + /** Data based on the type. */ + union + { + /** Number. */ + int64_t i64Num; + /** Flag. */ + bool fModule; + /** String - variable in size. */ + char aszString[1]; + } u; +} DBGDIGGERLINUXCFGITEM; +/** Pointer to a config database item. */ +typedef DBGDIGGERLINUXCFGITEM *PDBGDIGGERLINUXCFGITEM; +/** Pointer to a const config database item. */ +typedef const DBGDIGGERLINUXCFGITEM *PCDBGDIGGERLINUXCFGITEM; + +/** + * Linux guest OS digger instance data. + */ +typedef struct DBGDIGGERLINUX +{ + /** Whether the information is valid or not. + * (For fending off illegal interface method calls.) */ + bool fValid; + /** Set if 64-bit, clear if 32-bit. */ + bool f64Bit; + /** Set if the kallsyms table uses relative addressing, clear + * if absolute addresses are used. */ + bool fRelKrnlAddr; + /** The relative base when kernel symbols use offsets rather than + * absolute addresses. */ + RTGCUINTPTR uKernelRelativeBase; + /** The guest kernel version used for version comparisons. */ + uint32_t uKrnlVer; + /** The guest kernel major version. */ + uint32_t uKrnlVerMaj; + /** The guest kernel minor version. */ + uint32_t uKrnlVerMin; + /** The guest kernel build version. */ + uint32_t uKrnlVerBld; + + /** The address of the linux banner. + * This is set during probing. */ + DBGFADDRESS AddrLinuxBanner; + /** Kernel base address. + * This is set during probing, refined during kallsyms parsing. */ + DBGFADDRESS AddrKernelBase; + /** The kernel size. */ + uint32_t cbKernel; + + /** The number of kernel symbols (kallsyms_num_syms). + * This is set during init. */ + uint32_t cKernelSymbols; + /** The size of the kernel name table (sizeof(kallsyms_names)). */ + uint32_t cbKernelNames; + /** Number of entries in the kernel_markers table. */ + uint32_t cKernelNameMarkers; + /** The size of the kernel symbol token table. */ + uint32_t cbKernelTokenTable; + /** The address of the encoded kernel symbol names (kallsyms_names). */ + DBGFADDRESS AddrKernelNames; + /** The address of the kernel symbol addresses (kallsyms_addresses). */ + DBGFADDRESS AddrKernelAddresses; + /** The address of the kernel symbol name markers (kallsyms_markers). */ + DBGFADDRESS AddrKernelNameMarkers; + /** The address of the kernel symbol token table (kallsyms_token_table). */ + DBGFADDRESS AddrKernelTokenTable; + /** The address of the kernel symbol token index table (kallsyms_token_index). */ + DBGFADDRESS AddrKernelTokenIndex; + + /** The kernel message log interface. */ + DBGFOSIDMESG IDmesg; + + /** The config database root. */ + RTSTRSPACE hCfgDb; +} DBGDIGGERLINUX; +/** Pointer to the linux guest OS digger instance data. */ +typedef DBGDIGGERLINUX *PDBGDIGGERLINUX; + + +/** + * The current printk_log structure. + */ +typedef struct LNXPRINTKHDR +{ + /** Monotonic timestamp. */ + uint64_t nsTimestamp; + /** The total size of this message record. */ + uint16_t cbTotal; + /** The size of the text part (immediately follows the header). */ + uint16_t cbText; + /** The size of the optional dictionary part (follows the text). */ + uint16_t cbDict; + /** The syslog facility number. */ + uint8_t bFacility; + /** First 5 bits are internal flags, next 3 bits are log level. */ + uint8_t fFlagsAndLevel; +} LNXPRINTKHDR; +AssertCompileSize(LNXPRINTKHDR, 2*sizeof(uint64_t)); +/** Pointer to linux printk_log header. */ +typedef LNXPRINTKHDR *PLNXPRINTKHDR; +/** Pointer to linux const printk_log header. */ +typedef LNXPRINTKHDR const *PCLNXPRINTKHDR; + + +/********************************************************************************************************************************* +* Defined Constants And Macros * +*********************************************************************************************************************************/ +/** First kernel map address for 32bit Linux hosts (__START_KERNEL_map). */ +#define LNX32_KERNEL_ADDRESS_START UINT32_C(0xc0000000) +/** First kernel map address for 64bit Linux hosts (__START_KERNEL_map). */ +#define LNX64_KERNEL_ADDRESS_START UINT64_C(0xffffffff80000000) +/** Validates a 32-bit linux kernel address */ +#define LNX32_VALID_ADDRESS(Addr) ((Addr) > UINT32_C(0x80000000) && (Addr) < UINT32_C(0xfffff000)) +/** Validates a 64-bit linux kernel address */ +#define LNX64_VALID_ADDRESS(Addr) ((Addr) > UINT64_C(0xffff800000000000) && (Addr) < UINT64_C(0xfffffffffffff000)) + +/** The max kernel size. */ +#define LNX_MAX_KERNEL_SIZE UINT32_C(0x0f000000) +/** Maximum kernel log buffer size. */ +#define LNX_MAX_KERNEL_LOG_SIZE (16 * _1M) + +/** The maximum size we expect for kallsyms_names. */ +#define LNX_MAX_KALLSYMS_NAMES_SIZE UINT32_C(0x200000) +/** The maximum size we expect for kallsyms_token_table. */ +#define LNX_MAX_KALLSYMS_TOKEN_TABLE_SIZE UINT32_C(0x10000) +/** The minimum number of symbols we expect in kallsyms_num_syms. */ +#define LNX_MIN_KALLSYMS_SYMBOLS UINT32_C(2048) +/** The maximum number of symbols we expect in kallsyms_num_syms. */ +#define LNX_MAX_KALLSYMS_SYMBOLS UINT32_C(1048576) +/** The min length an encoded symbol in kallsyms_names is expected to have. */ +#define LNX_MIN_KALLSYMS_ENC_LENGTH UINT8_C(1) +/** The max length an encoded symbol in kallsyms_names is expected to have. + * @todo check real life here. */ +#define LNX_MAX_KALLSYMS_ENC_LENGTH UINT8_C(28) +/** The approximate maximum length of a string token. */ +#define LNX_MAX_KALLSYMS_TOKEN_LEN UINT16_C(32) +/** Maximum compressed config size expected. */ +#define LNX_MAX_COMPRESSED_CFG_SIZE _1M + +/** Module tag for linux ('linuxmod' on little endian ASCII systems). */ +#define DIG_LNX_MOD_TAG UINT64_C(0x545f5d78758e898c) +/** Macro for building a Linux kernel version which can be used for comparisons. */ +#define LNX_MK_VER(major, minor, build) (((major) << 22) | ((minor) << 12) | (build)) + + +/********************************************************************************************************************************* +* Internal Functions * +*********************************************************************************************************************************/ +static DECLCALLBACK(int) dbgDiggerLinuxInit(PUVM pUVM, PCVMMR3VTABLE pVMM, void *pvData); + + +/********************************************************************************************************************************* +* Global Variables * +*********************************************************************************************************************************/ +/** Table of common linux kernel addresses. */ +static uint64_t g_au64LnxKernelAddresses[] = +{ + UINT64_C(0xc0100000), + UINT64_C(0x90100000), + UINT64_C(0xffffffff80200000) +}; + +static const uint8_t g_abLinuxVersion[] = "Linux version "; +/** The needle for searching for the kernel log area (the value is observed in pretty much all 32bit and 64bit x86 kernels). + * This needle should appear only once in the memory due to the address being filled in by a format string. */ +static const uint8_t g_abKrnlLogNeedle[] = "BIOS-e820: [mem 0x0000000000000000"; + + +/** + * Tries to resolve the kernel log buffer start and end by searching for needle. + * + * @returns VBox status code. + * @param pThis The Linux digger data. + * @param pUVM The VM handle. + * @param pVMM The VMM function table. + * @param pGCPtrLogBuf Where to store the start of the kernel log buffer on success. + * @param pcbLogBuf Where to store the size of the kernel log buffer on success. + */ +static int dbgDiggerLinuxKrnlLogBufFindByNeedle(PDBGDIGGERLINUX pThis, PUVM pUVM, PCVMMR3VTABLE pVMM, + RTGCPTR *pGCPtrLogBuf, uint32_t *pcbLogBuf) +{ + int rc = VINF_SUCCESS; + + /* Try to find the needle, it should be very early in the kernel log buffer. */ + DBGFADDRESS AddrScan; + DBGFADDRESS AddrHit; + pVMM->pfnDBGFR3AddrFromFlat(pUVM, &AddrScan, pThis->f64Bit ? LNX64_KERNEL_ADDRESS_START : LNX32_KERNEL_ADDRESS_START); + + rc = pVMM->pfnDBGFR3MemScan(pUVM, 0 /*idCpu*/, &AddrScan, ~(RTGCUINTPTR)0, 1 /*uAlign*/, + g_abKrnlLogNeedle, sizeof(g_abKrnlLogNeedle) - 1, &AddrHit); + if (RT_SUCCESS(rc)) + { + uint32_t cbLogBuf = 0; + uint64_t tsLastNs = 0; + DBGFADDRESS AddrCur; + + pVMM->pfnDBGFR3AddrSub(&AddrHit, sizeof(LNXPRINTKHDR)); + AddrCur = AddrHit; + + /* Try to find the end of the kernel log buffer. */ + for (;;) + { + if (cbLogBuf >= LNX_MAX_KERNEL_LOG_SIZE) + break; + + LNXPRINTKHDR Hdr; + rc = pVMM->pfnDBGFR3MemRead(pUVM, 0 /*idCpu*/, &AddrCur, &Hdr, sizeof(Hdr)); + if (RT_SUCCESS(rc)) + { + uint32_t const cbLogAlign = 4; + + /* + * If the header does not look valid anymore we stop. + * Timestamps are monotonically increasing. + */ + if ( !Hdr.cbTotal /* Zero entry size means there is no record anymore, doesn't make sense to look futher. */ + || Hdr.cbText + Hdr.cbDict + sizeof(Hdr) > Hdr.cbTotal + || (Hdr.cbTotal & (cbLogAlign - 1)) != 0 + || tsLastNs > Hdr.nsTimestamp) + break; + + /** @todo Maybe read text part and verify it is all ASCII. */ + + cbLogBuf += Hdr.cbTotal; + pVMM->pfnDBGFR3AddrAdd(&AddrCur, Hdr.cbTotal); + } + + if (RT_FAILURE(rc)) + break; + } + + /** @todo Go back to find the start address of the kernel log (or we loose potential kernel log messages). */ + + if ( RT_SUCCESS(rc) + && cbLogBuf) + { + /* Align log buffer size to a power of two. */ + uint32_t idxBitLast = ASMBitLastSetU32(cbLogBuf); + idxBitLast--; /* There is at least one bit set, see check above. */ + + if (cbLogBuf & (RT_BIT_32(idxBitLast) - 1)) + idxBitLast++; + + *pGCPtrLogBuf = AddrHit.FlatPtr; + *pcbLogBuf = RT_MIN(RT_BIT_32(idxBitLast), LNX_MAX_KERNEL_LOG_SIZE); + } + else if (RT_SUCCESS(rc)) + rc = VERR_NOT_FOUND; + } + + return rc; +} + + +/** + * Converts a given offset into an absolute address if relative kernel offsets are used for + * kallsyms. + * + * @returns The absolute kernel address. + * @param pThis The Linux digger data. + * @param uOffset The offset to convert. + */ +DECLINLINE(RTGCUINTPTR) dbgDiggerLinuxConvOffsetToAddr(PDBGDIGGERLINUX pThis, int32_t uOffset) +{ + RTGCUINTPTR uAddr; + + /* + * How the absolute address is calculated from the offset depends on the + * CONFIG_KALLSYMS_ABSOLUTE_PERCPU config which is only set for 64bit + * SMP kernels (we assume that all 64bit kernels always have SMP enabled too). + */ + if (pThis->f64Bit) + { + if (uOffset >= 0) + uAddr = uOffset; + else + uAddr = pThis->uKernelRelativeBase - 1 - uOffset; + } + else + uAddr = pThis->uKernelRelativeBase + (uint32_t)uOffset; + + return uAddr; +} + +/** + * Disassembles a simple getter returning the value for it. + * + * @returns VBox status code. + * @param pThis The Linux digger data. + * @param pUVM The VM handle. + * @param pVMM The VMM function table. + * @param hMod The module to use. + * @param pszSymbol The symbol of the getter. + * @param pvVal Where to store the value on success. + * @param cbVal Size of the value in bytes. + */ +static int dbgDiggerLinuxDisassembleSimpleGetter(PDBGDIGGERLINUX pThis, PUVM pUVM, PCVMMR3VTABLE pVMM, RTDBGMOD hMod, + const char *pszSymbol, void *pvVal, uint32_t cbVal) +{ + int rc = VINF_SUCCESS; + + RTDBGSYMBOL SymInfo; + rc = RTDbgModSymbolByName(hMod, pszSymbol, &SymInfo); + if (RT_SUCCESS(rc)) + { + /* + * Do the diassembling. Disassemble until a ret instruction is encountered + * or a limit is reached (don't want to disassemble for too long as the getter + * should be short). + * push and pop instructions are skipped as well as any mov instructions not + * touching the rax or eax register (depending on the size of the value). + */ + unsigned cInstrDisassembled = 0; + uint32_t offInstr = 0; + bool fRet = false; + DISSTATE DisState; + RT_ZERO(DisState); + + do + { + DBGFADDRESS Addr; + RTGCPTR GCPtrCur = (RTGCPTR)SymInfo.Value + pThis->AddrKernelBase.FlatPtr + offInstr; + pVMM->pfnDBGFR3AddrFromFlat(pUVM, &Addr, GCPtrCur); + + /* Prefetch the instruction. */ + uint8_t abInstr[32]; + rc = pVMM->pfnDBGFR3MemRead(pUVM, 0 /*idCpu*/, &Addr, &abInstr[0], sizeof(abInstr)); + if (RT_SUCCESS(rc)) + { + uint32_t cbInstr = 0; + + rc = DISInstr(&abInstr[0], pThis->f64Bit ? DISCPUMODE_64BIT : DISCPUMODE_32BIT, &DisState, &cbInstr); + if (RT_SUCCESS(rc)) + { + switch (DisState.pCurInstr->uOpcode) + { + case OP_PUSH: + case OP_POP: + case OP_NOP: + case OP_LEA: + break; + case OP_RETN: + /* Getter returned, abort disassembling. */ + fRet = true; + break; + case OP_MOV: + /* + * Check that the destination is either rax or eax depending on the + * value size. + * + * Param1 is the destination and Param2 the source. + */ + if ( ( ( (DisState.Param1.fUse & (DISUSE_BASE | DISUSE_REG_GEN32)) + && cbVal == sizeof(uint32_t)) + || ( (DisState.Param1.fUse & (DISUSE_BASE | DISUSE_REG_GEN64)) + && cbVal == sizeof(uint64_t))) + && DisState.Param1.Base.idxGenReg == DISGREG_RAX) + { + /* Parse the source. */ + if (DisState.Param2.fUse & (DISUSE_IMMEDIATE32 | DISUSE_IMMEDIATE64)) + memcpy(pvVal, &DisState.Param2.uValue, cbVal); + else if (DisState.Param2.fUse & (DISUSE_RIPDISPLACEMENT32|DISUSE_DISPLACEMENT32|DISUSE_DISPLACEMENT64)) + { + RTGCPTR GCPtrVal = 0; + + if (DisState.Param2.fUse & DISUSE_RIPDISPLACEMENT32) + GCPtrVal = GCPtrCur + DisState.Param2.uDisp.i32 + cbInstr; + else if (DisState.Param2.fUse & DISUSE_DISPLACEMENT32) + GCPtrVal = (RTGCPTR)DisState.Param2.uDisp.u32; + else if (DisState.Param2.fUse & DISUSE_DISPLACEMENT64) + GCPtrVal = (RTGCPTR)DisState.Param2.uDisp.u64; + else + AssertMsgFailedBreakStmt(("Invalid displacement\n"), rc = VERR_INVALID_STATE); + + DBGFADDRESS AddrVal; + rc = pVMM->pfnDBGFR3MemRead(pUVM, 0 /*idCpu*/, + pVMM->pfnDBGFR3AddrFromFlat(pUVM, &AddrVal, GCPtrVal), + pvVal, cbVal); + } + } + break; + default: + /* All other instructions will cause an error for now (playing safe here). */ + rc = VERR_INVALID_PARAMETER; + break; + } + cInstrDisassembled++; + offInstr += cbInstr; + } + } + } while ( RT_SUCCESS(rc) + && cInstrDisassembled < 20 + && !fRet); + } + + return rc; +} + +/** + * Try to get at the log buffer starting address and size by disassembling emit_log_char. + * + * @returns VBox status code. + * @param pThis The Linux digger data. + * @param pUVM The VM handle. + * @param pVMM The VMM function table. + * @param hMod The module to use. + * @param pGCPtrLogBuf Where to store the log buffer pointer on success. + * @param pcbLogBuf Where to store the size of the log buffer on success. + */ +static int dbgDiggerLinuxQueryAsciiLogBufferPtrs(PDBGDIGGERLINUX pThis, PUVM pUVM, PCVMMR3VTABLE pVMM, RTDBGMOD hMod, + RTGCPTR *pGCPtrLogBuf, uint32_t *pcbLogBuf) +{ + int rc = VINF_SUCCESS; + + /** + * We disassemble emit_log_char to get at the log buffer address and size. + * This is used in case the symbols are not exported in kallsyms. + * + * This is what it typically looks like: + * vmlinux!emit_log_char: + * %00000000c01204a1 56 push esi + * %00000000c01204a2 8b 35 d0 1c 34 c0 mov esi, dword [0c0341cd0h] + * %00000000c01204a8 53 push ebx + * %00000000c01204a9 8b 1d 74 3b 3e c0 mov ebx, dword [0c03e3b74h] + * %00000000c01204af 8b 0d d8 1c 34 c0 mov ecx, dword [0c0341cd8h] + * %00000000c01204b5 8d 56 ff lea edx, [esi-001h] + * %00000000c01204b8 21 da and edx, ebx + * %00000000c01204ba 88 04 11 mov byte [ecx+edx], al + * %00000000c01204bd 8d 53 01 lea edx, [ebx+001h] + * %00000000c01204c0 89 d0 mov eax, edx + * [...] + */ + RTDBGSYMBOL SymInfo; + rc = RTDbgModSymbolByName(hMod, "emit_log_char", &SymInfo); + if (RT_SUCCESS(rc)) + { + /* + * Do the diassembling. Disassemble until a ret instruction is encountered + * or a limit is reached (don't want to disassemble for too long as the getter + * should be short). Certain instructions found are ignored (push, nop, etc.). + */ + unsigned cInstrDisassembled = 0; + uint32_t offInstr = 0; + bool fRet = false; + DISSTATE DisState; + unsigned cAddressesUsed = 0; + struct { size_t cb; RTGCPTR GCPtrOrigSrc; } aAddresses[5]; + RT_ZERO(DisState); + RT_ZERO(aAddresses); + + do + { + DBGFADDRESS Addr; + RTGCPTR GCPtrCur = (RTGCPTR)SymInfo.Value + pThis->AddrKernelBase.FlatPtr + offInstr; + pVMM->pfnDBGFR3AddrFromFlat(pUVM, &Addr, GCPtrCur); + + /* Prefetch the instruction. */ + uint8_t abInstr[32]; + rc = pVMM->pfnDBGFR3MemRead(pUVM, 0 /*idCpu*/, &Addr, &abInstr[0], sizeof(abInstr)); + if (RT_SUCCESS(rc)) + { + uint32_t cbInstr = 0; + + rc = DISInstr(&abInstr[0], pThis->f64Bit ? DISCPUMODE_64BIT : DISCPUMODE_32BIT, &DisState, &cbInstr); + if (RT_SUCCESS(rc)) + { + switch (DisState.pCurInstr->uOpcode) + { + case OP_PUSH: + case OP_POP: + case OP_NOP: + case OP_LEA: + case OP_AND: + case OP_CBW: + case OP_DEC: + break; + case OP_RETN: + /* emit_log_char returned, abort disassembling. */ + rc = VERR_NOT_FOUND; + fRet = true; + break; + case OP_MOV: + case OP_MOVSXD: + /* + * If a mov is encountered writing to memory with al (or dil for amd64) being the source the + * character is stored and we can infer the base address and size of the log buffer from + * the source addresses. + */ + if ( (DisState.Param2.fUse & DISUSE_REG_GEN8) + && ( (DisState.Param2.Base.idxGenReg == DISGREG_AL && !pThis->f64Bit) + || (DisState.Param2.Base.idxGenReg == DISGREG_DIL && pThis->f64Bit)) + && DISUSE_IS_EFFECTIVE_ADDR(DisState.Param1.fUse)) + { + RTGCPTR GCPtrLogBuf = 0; + uint32_t cbLogBuf = 0; + + /* + * We can stop disassembling now and inspect all registers, look for a valid kernel address first. + * Only one of the accessed registers should hold a valid kernel address. + * For the log size look for the biggest non kernel address. + */ + for (unsigned i = 0; i < cAddressesUsed; i++) + { + DBGFADDRESS AddrVal; + union { uint8_t abVal[8]; uint32_t u32Val; uint64_t u64Val; } Val; + + rc = pVMM->pfnDBGFR3MemRead(pUVM, 0 /*idCpu*/, + pVMM->pfnDBGFR3AddrFromFlat(pUVM, &AddrVal, + aAddresses[i].GCPtrOrigSrc), + &Val.abVal[0], aAddresses[i].cb); + if (RT_SUCCESS(rc)) + { + if (pThis->f64Bit && aAddresses[i].cb == sizeof(uint64_t)) + { + if (LNX64_VALID_ADDRESS(Val.u64Val)) + { + if (GCPtrLogBuf == 0) + GCPtrLogBuf = Val.u64Val; + else + { + rc = VERR_NOT_FOUND; + break; + } + } + } + else + { + AssertMsgBreakStmt(aAddresses[i].cb == sizeof(uint32_t), + ("Invalid value size\n"), rc = VERR_INVALID_STATE); + + /* Might be a kernel address or a size indicator. */ + if (!pThis->f64Bit && LNX32_VALID_ADDRESS(Val.u32Val)) + { + if (GCPtrLogBuf == 0) + GCPtrLogBuf = Val.u32Val; + else + { + rc = VERR_NOT_FOUND; + break; + } + } + else + { + /* + * The highest value will be the log buffer because the other + * accessed variables are indexes into the buffer and hence + * always smaller than the size. + */ + if (cbLogBuf < Val.u32Val) + cbLogBuf = Val.u32Val; + } + } + } + } + + if ( RT_SUCCESS(rc) + && GCPtrLogBuf != 0 + && cbLogBuf != 0) + { + *pGCPtrLogBuf = GCPtrLogBuf; + *pcbLogBuf = cbLogBuf; + } + else if (RT_SUCCESS(rc)) + rc = VERR_NOT_FOUND; + + fRet = true; + break; + } + else + { + /* + * In case of a memory to register move store the destination register index and the + * source address in the relation table for later processing. + */ + if ( (DisState.Param1.fUse & (DISUSE_BASE | DISUSE_REG_GEN32 | DISUSE_REG_GEN64)) + && (DisState.Param2.cb == sizeof(uint32_t) || DisState.Param2.cb == sizeof(uint64_t)) + && (DisState.Param2.fUse & (DISUSE_RIPDISPLACEMENT32|DISUSE_DISPLACEMENT32|DISUSE_DISPLACEMENT64))) + { + RTGCPTR GCPtrVal = 0; + + if (DisState.Param2.fUse & DISUSE_RIPDISPLACEMENT32) + GCPtrVal = GCPtrCur + DisState.Param2.uDisp.i32 + cbInstr; + else if (DisState.Param2.fUse & DISUSE_DISPLACEMENT32) + GCPtrVal = (RTGCPTR)DisState.Param2.uDisp.u32; + else if (DisState.Param2.fUse & DISUSE_DISPLACEMENT64) + GCPtrVal = (RTGCPTR)DisState.Param2.uDisp.u64; + else + AssertMsgFailedBreakStmt(("Invalid displacement\n"), rc = VERR_INVALID_STATE); + + if (cAddressesUsed < RT_ELEMENTS(aAddresses)) + { + /* movsxd reads always 32bits. */ + if (DisState.pCurInstr->uOpcode == OP_MOVSXD) + aAddresses[cAddressesUsed].cb = sizeof(uint32_t); + else + aAddresses[cAddressesUsed].cb = DisState.Param2.cb; + aAddresses[cAddressesUsed].GCPtrOrigSrc = GCPtrVal; + cAddressesUsed++; + } + else + { + rc = VERR_INVALID_PARAMETER; + break; + } + } + } + break; + default: + /* All other instructions will cause an error for now (playing safe here). */ + rc = VERR_INVALID_PARAMETER; + break; + } + cInstrDisassembled++; + offInstr += cbInstr; + } + } + } while ( RT_SUCCESS(rc) + && cInstrDisassembled < 20 + && !fRet); + } + + return rc; +} + +/** + * Try to get at the log buffer starting address and size by disassembling some exposed helpers. + * + * @returns VBox status code. + * @param pThis The Linux digger data. + * @param pUVM The VM handle. + * @param pVMM The VMM function table. + * @param hMod The module to use. + * @param pGCPtrLogBuf Where to store the log buffer pointer on success. + * @param pcbLogBuf Where to store the size of the log buffer on success. + */ +static int dbgDiggerLinuxQueryLogBufferPtrs(PDBGDIGGERLINUX pThis, PUVM pUVM, PCVMMR3VTABLE pVMM, RTDBGMOD hMod, + RTGCPTR *pGCPtrLogBuf, uint32_t *pcbLogBuf) +{ + int rc = VINF_SUCCESS; + + struct { void *pvVar; uint32_t cbHost, cbGuest; const char *pszSymbol; } aSymbols[] = + { + { pGCPtrLogBuf, (uint32_t)sizeof(RTGCPTR), (uint32_t)(pThis->f64Bit ? sizeof(uint64_t) : sizeof(uint32_t)), "log_buf_addr_get" }, + { pcbLogBuf, (uint32_t)sizeof(uint32_t), (uint32_t)sizeof(uint32_t), "log_buf_len_get" } + }; + for (uint32_t i = 0; i < RT_ELEMENTS(aSymbols) && RT_SUCCESS(rc); i++) + { + RT_BZERO(aSymbols[i].pvVar, aSymbols[i].cbHost); + Assert(aSymbols[i].cbHost >= aSymbols[i].cbGuest); + rc = dbgDiggerLinuxDisassembleSimpleGetter(pThis, pUVM, pVMM, hMod, aSymbols[i].pszSymbol, + aSymbols[i].pvVar, aSymbols[i].cbGuest); + } + + return rc; +} + +/** + * Returns whether the log buffer is a simple ascii buffer or a record based implementation + * based on the kernel version found. + * + * @returns Flag whether the log buffer is the simple ascii buffer. + * @param pThis The Linux digger data. + * @param pUVM The user mode VM handle. + * @param pVMM The VMM function table. + */ +static bool dbgDiggerLinuxLogBufferIsAsciiBuffer(PDBGDIGGERLINUX pThis, PUVM pUVM, PCVMMR3VTABLE pVMM) +{ + char szTmp[128]; + char const *pszVer = &szTmp[sizeof(g_abLinuxVersion) - 1]; + + RT_ZERO(szTmp); + int rc = pVMM->pfnDBGFR3MemReadString(pUVM, 0, &pThis->AddrLinuxBanner, szTmp, sizeof(szTmp) - 1); + if ( RT_SUCCESS(rc) + && RTStrVersionCompare(pszVer, "3.4") == -1) + return true; + + return false; +} + +/** + * Worker to get at the kernel log for pre 3.4 kernels where the log buffer was just a char buffer. + * + * @returns VBox status code. + * @param pThis The Linux digger data. + * @param pUVM The VM user mdoe handle. + * @param pVMM The VMM function table. + * @param hMod The debug module handle. + * @param fFlags Flags reserved for future use, MBZ. + * @param cMessages The number of messages to retrieve, counting from the + * end of the log (i.e. like tail), use UINT32_MAX for all. + * @param pszBuf The output buffer. + * @param cbBuf The buffer size. + * @param pcbActual Where to store the number of bytes actually returned, + * including zero terminator. On VERR_BUFFER_OVERFLOW this + * holds the necessary buffer size. Optional. + */ +static int dbgDiggerLinuxLogBufferQueryAscii(PDBGDIGGERLINUX pThis, PUVM pUVM, PCVMMR3VTABLE pVMM, RTDBGMOD hMod, + uint32_t fFlags, uint32_t cMessages, + char *pszBuf, size_t cbBuf, size_t *pcbActual) +{ + RT_NOREF2(fFlags, cMessages); + int rc = VINF_SUCCESS; + RTGCPTR GCPtrLogBuf; + uint32_t cbLogBuf; + + struct { void *pvVar; size_t cbHost, cbGuest; const char *pszSymbol; } aSymbols[] = + { + { &GCPtrLogBuf, sizeof(GCPtrLogBuf), pThis->f64Bit ? sizeof(uint64_t) : sizeof(uint32_t), "log_buf" }, + { &cbLogBuf, sizeof(cbLogBuf), sizeof(cbLogBuf), "log_buf_len" }, + }; + for (uint32_t i = 0; i < RT_ELEMENTS(aSymbols); i++) + { + RTDBGSYMBOL SymInfo; + rc = RTDbgModSymbolByName(hMod, aSymbols[i].pszSymbol, &SymInfo); + if (RT_SUCCESS(rc)) + { + RT_BZERO(aSymbols[i].pvVar, aSymbols[i].cbHost); + Assert(aSymbols[i].cbHost >= aSymbols[i].cbGuest); + DBGFADDRESS Addr; + rc = pVMM->pfnDBGFR3MemRead(pUVM, 0 /*idCpu*/, + pVMM->pfnDBGFR3AddrFromFlat(pUVM, &Addr, + (RTGCPTR)SymInfo.Value + pThis->AddrKernelBase.FlatPtr), + aSymbols[i].pvVar, aSymbols[i].cbGuest); + if (RT_SUCCESS(rc)) + continue; + LogRel(("dbgDiggerLinuxIDmsg_QueryKernelLog: Reading '%s' at %RGv: %Rrc\n", aSymbols[i].pszSymbol, Addr.FlatPtr, rc)); + } + else + LogRel(("dbgDiggerLinuxIDmsg_QueryKernelLog: Error looking up '%s': %Rrc\n", aSymbols[i].pszSymbol, rc)); + rc = VERR_NOT_FOUND; + break; + } + + /* + * Some kernels don't expose the variables in kallsyms so we have to try disassemble + * some public helpers to get at the addresses. + * + * @todo: Maybe cache those values so we don't have to do the heavy work every time? + */ + if (rc == VERR_NOT_FOUND) + { + rc = dbgDiggerLinuxQueryAsciiLogBufferPtrs(pThis, pUVM, pVMM, hMod, &GCPtrLogBuf, &cbLogBuf); + if (RT_FAILURE(rc)) + return rc; + } + + /* + * Check if the values make sense. + */ + if (pThis->f64Bit ? !LNX64_VALID_ADDRESS(GCPtrLogBuf) : !LNX32_VALID_ADDRESS(GCPtrLogBuf)) + { + LogRel(("dbgDiggerLinuxIDmsg_QueryKernelLog: 'log_buf' value %RGv is not valid.\n", GCPtrLogBuf)); + return VERR_NOT_FOUND; + } + if ( cbLogBuf < 4096 + || !RT_IS_POWER_OF_TWO(cbLogBuf) + || cbLogBuf > 16*_1M) + { + LogRel(("dbgDiggerLinuxIDmsg_QueryKernelLog: 'log_buf_len' value %#x is not valid.\n", cbLogBuf)); + return VERR_NOT_FOUND; + } + + /* + * Read the whole log buffer. + */ + uint8_t *pbLogBuf = (uint8_t *)RTMemAlloc(cbLogBuf); + if (!pbLogBuf) + { + LogRel(("dbgDiggerLinuxIDmsg_QueryKernelLog: Failed to allocate %#x bytes for log buffer\n", cbLogBuf)); + return VERR_NO_MEMORY; + } + DBGFADDRESS Addr; + rc = pVMM->pfnDBGFR3MemRead(pUVM, 0 /*idCpu*/, pVMM->pfnDBGFR3AddrFromFlat(pUVM, &Addr, GCPtrLogBuf), pbLogBuf, cbLogBuf); + if (RT_FAILURE(rc)) + { + LogRel(("dbgDiggerLinuxIDmsg_QueryKernelLog: Error reading %#x bytes of log buffer at %RGv: %Rrc\n", + cbLogBuf, Addr.FlatPtr, rc)); + RTMemFree(pbLogBuf); + return VERR_NOT_FOUND; + } + + /** @todo Try to parse where the single messages start to make use of cMessages. */ + size_t cchLength = RTStrNLen((const char *)pbLogBuf, cbLogBuf); + memcpy(&pszBuf[0], pbLogBuf, RT_MIN(cbBuf, cchLength)); + + /* Done with the buffer. */ + RTMemFree(pbLogBuf); + + /* Set return size value. */ + if (pcbActual) + *pcbActual = RT_MIN(cbBuf, cchLength); + + return cbBuf <= cchLength ? VERR_BUFFER_OVERFLOW : VINF_SUCCESS; +} + + +/** + * Worker to process a given record based kernel log. + * + * @returns VBox status code. + * @param pThis The Linux digger data. + * @param pUVM The VM user mode handle. + * @param pVMM The VMM function table. + * @param GCPtrLogBuf Flat guest address of the start of the log buffer. + * @param cbLogBuf Power of two aligned size of the log buffer. + * @param idxFirst Index in the log bfufer of the first message. + * @param idxNext Index where to write hte next message in the log buffer. + * @param fFlags Flags reserved for future use, MBZ. + * @param cMessages The number of messages to retrieve, counting from the + * end of the log (i.e. like tail), use UINT32_MAX for all. + * @param pszBuf The output buffer. + * @param cbBuf The buffer size. + * @param pcbActual Where to store the number of bytes actually returned, + * including zero terminator. On VERR_BUFFER_OVERFLOW this + * holds the necessary buffer size. Optional. + */ +static int dbgDiggerLinuxKrnLogBufferProcess(PDBGDIGGERLINUX pThis, PUVM pUVM, PCVMMR3VTABLE pVMM, RTGCPTR GCPtrLogBuf, + uint32_t cbLogBuf, uint32_t idxFirst, uint32_t idxNext, + uint32_t fFlags, uint32_t cMessages, char *pszBuf, size_t cbBuf, + size_t *pcbActual) +{ + RT_NOREF(fFlags); + + /* + * Check if the values make sense. + */ + if (pThis->f64Bit ? !LNX64_VALID_ADDRESS(GCPtrLogBuf) : !LNX32_VALID_ADDRESS(GCPtrLogBuf)) + { + LogRel(("dbgDiggerLinuxIDmsg_QueryKernelLog: 'log_buf' value %RGv is not valid.\n", GCPtrLogBuf)); + return VERR_NOT_FOUND; + } + if ( cbLogBuf < _4K + || !RT_IS_POWER_OF_TWO(cbLogBuf) + || cbLogBuf > LNX_MAX_KERNEL_LOG_SIZE) + { + LogRel(("dbgDiggerLinuxIDmsg_QueryKernelLog: 'log_buf_len' value %#x is not valid.\n", cbLogBuf)); + return VERR_NOT_FOUND; + } + uint32_t const cbLogAlign = 4; + if ( idxFirst > cbLogBuf - sizeof(LNXPRINTKHDR) + || (idxFirst & (cbLogAlign - 1)) != 0) + { + LogRel(("dbgDiggerLinuxIDmsg_QueryKernelLog: 'log_first_idx' value %#x is not valid.\n", idxFirst)); + return VERR_NOT_FOUND; + } + if ( idxNext > cbLogBuf - sizeof(LNXPRINTKHDR) + || (idxNext & (cbLogAlign - 1)) != 0) + { + LogRel(("dbgDiggerLinuxIDmsg_QueryKernelLog: 'log_next_idx' value %#x is not valid.\n", idxNext)); + return VERR_NOT_FOUND; + } + + /* + * Read the whole log buffer. + */ + uint8_t *pbLogBuf = (uint8_t *)RTMemAlloc(cbLogBuf); + if (!pbLogBuf) + { + LogRel(("dbgDiggerLinuxIDmsg_QueryKernelLog: Failed to allocate %#x bytes for log buffer\n", cbLogBuf)); + return VERR_NO_MEMORY; + } + DBGFADDRESS Addr; + int rc = pVMM->pfnDBGFR3MemRead(pUVM, 0 /*idCpu*/, pVMM->pfnDBGFR3AddrFromFlat(pUVM, &Addr, GCPtrLogBuf), pbLogBuf, cbLogBuf); + if (RT_FAILURE(rc)) + { + LogRel(("dbgDiggerLinuxIDmsg_QueryKernelLog: Error reading %#x bytes of log buffer at %RGv: %Rrc\n", + cbLogBuf, Addr.FlatPtr, rc)); + RTMemFree(pbLogBuf); + return VERR_NOT_FOUND; + } + + /* + * Count the messages in the buffer while doing some basic validation. + */ + uint32_t const cbUsed = idxFirst == idxNext ? cbLogBuf /* could be empty... */ + : idxFirst < idxNext ? idxNext - idxFirst : cbLogBuf - idxFirst + idxNext; + uint32_t cbLeft = cbUsed; + uint32_t offCur = idxFirst; + uint32_t cLogMsgs = 0; + + while (cbLeft > 0) + { + PCLNXPRINTKHDR pHdr = (PCLNXPRINTKHDR)&pbLogBuf[offCur]; + if (!pHdr->cbTotal) + { + /* Wrap around packet, most likely... */ + if (cbLogBuf - offCur >= cbLeft) + break; + offCur = 0; + pHdr = (PCLNXPRINTKHDR)&pbLogBuf[offCur]; + } + if (RT_UNLIKELY( pHdr->cbTotal > cbLogBuf - sizeof(*pHdr) - offCur + || pHdr->cbTotal > cbLeft + || (pHdr->cbTotal & (cbLogAlign - 1)) != 0 + || pHdr->cbTotal < (uint32_t)pHdr->cbText + (uint32_t)pHdr->cbDict + sizeof(*pHdr) )) + { + LogRel(("dbgDiggerLinuxIDmsg_QueryKernelLog: Invalid printk_log record at %#x: cbTotal=%#x cbText=%#x cbDict=%#x cbLogBuf=%#x cbLeft=%#x\n", + offCur, pHdr->cbTotal, pHdr->cbText, pHdr->cbDict, cbLogBuf, cbLeft)); + break; + } + + if (pHdr->cbText > 0) + cLogMsgs++; + + /* next */ + offCur += pHdr->cbTotal; + cbLeft -= pHdr->cbTotal; + } + if (!cLogMsgs) + { + RTMemFree(pbLogBuf); + return VERR_NOT_FOUND; + } + + /* + * Copy the messages into the output buffer. + */ + offCur = idxFirst; + cbLeft = cbUsed - cbLeft; + + /* Skip messages that the caller doesn't want. */ + if (cMessages < cLogMsgs) + { + uint32_t cToSkip = cLogMsgs - cMessages; + cLogMsgs -= cToSkip; + + while (cToSkip > 0) + { + PCLNXPRINTKHDR pHdr = (PCLNXPRINTKHDR)&pbLogBuf[offCur]; + if (!pHdr->cbTotal) + { + offCur = 0; + pHdr = (PCLNXPRINTKHDR)&pbLogBuf[offCur]; + } + if (pHdr->cbText > 0) + cToSkip--; + + /* next */ + offCur += pHdr->cbTotal; + cbLeft -= pHdr->cbTotal; + } + } + + /* Now copy the messages. */ + size_t offDst = 0; + while (cbLeft > 0) + { + PCLNXPRINTKHDR pHdr = (PCLNXPRINTKHDR)&pbLogBuf[offCur]; + if ( !pHdr->cbTotal + || !cLogMsgs) + { + if (cbLogBuf - offCur >= cbLeft) + break; + offCur = 0; + pHdr = (PCLNXPRINTKHDR)&pbLogBuf[offCur]; + } + + if (pHdr->cbText > 0) + { + char *pchText = (char *)(pHdr + 1); + size_t cchText = RTStrNLen(pchText, pHdr->cbText); + if (offDst + cchText < cbBuf) + { + memcpy(&pszBuf[offDst], pHdr + 1, cchText); + pszBuf[offDst + cchText] = '\n'; + } + else if (offDst < cbBuf) + memcpy(&pszBuf[offDst], pHdr + 1, cbBuf - offDst); + offDst += cchText + 1; + } + + /* next */ + offCur += pHdr->cbTotal; + cbLeft -= pHdr->cbTotal; + } + + /* Done with the buffer. */ + RTMemFree(pbLogBuf); + + /* Make sure we've reserved a char for the terminator. */ + if (!offDst) + offDst = 1; + + /* Set return size value. */ + if (pcbActual) + *pcbActual = offDst; + + if (offDst <= cbBuf) + return VINF_SUCCESS; + return VERR_BUFFER_OVERFLOW; +} + + +/** + * Worker to get at the kernel log for post 3.4 kernels where the log buffer contains records. + * + * @returns VBox status code. + * @param pThis The Linux digger data. + * @param pUVM The VM user mdoe handle. + * @param pVMM The VMM function table. + * @param hMod The debug module handle. + * @param fFlags Flags reserved for future use, MBZ. + * @param cMessages The number of messages to retrieve, counting from the + * end of the log (i.e. like tail), use UINT32_MAX for all. + * @param pszBuf The output buffer. + * @param cbBuf The buffer size. + * @param pcbActual Where to store the number of bytes actually returned, + * including zero terminator. On VERR_BUFFER_OVERFLOW this + * holds the necessary buffer size. Optional. + */ +static int dbgDiggerLinuxLogBufferQueryRecords(PDBGDIGGERLINUX pThis, PUVM pUVM, PCVMMR3VTABLE pVMM, RTDBGMOD hMod, + uint32_t fFlags, uint32_t cMessages, + char *pszBuf, size_t cbBuf, size_t *pcbActual) +{ + int rc = VINF_SUCCESS; + RTGCPTR GCPtrLogBuf; + uint32_t cbLogBuf; + uint32_t idxFirst; + uint32_t idxNext; + + struct { void *pvVar; size_t cbHost, cbGuest; const char *pszSymbol; } aSymbols[] = + { + { &GCPtrLogBuf, sizeof(GCPtrLogBuf), pThis->f64Bit ? sizeof(uint64_t) : sizeof(uint32_t), "log_buf" }, + { &cbLogBuf, sizeof(cbLogBuf), sizeof(cbLogBuf), "log_buf_len" }, + { &idxFirst, sizeof(idxFirst), sizeof(idxFirst), "log_first_idx" }, + { &idxNext, sizeof(idxNext), sizeof(idxNext), "log_next_idx" }, + }; + for (uint32_t i = 0; i < RT_ELEMENTS(aSymbols); i++) + { + RTDBGSYMBOL SymInfo; + rc = RTDbgModSymbolByName(hMod, aSymbols[i].pszSymbol, &SymInfo); + if (RT_SUCCESS(rc)) + { + RT_BZERO(aSymbols[i].pvVar, aSymbols[i].cbHost); + Assert(aSymbols[i].cbHost >= aSymbols[i].cbGuest); + DBGFADDRESS Addr; + rc = pVMM->pfnDBGFR3MemRead(pUVM, 0 /*idCpu*/, + pVMM->pfnDBGFR3AddrFromFlat(pUVM, &Addr, + (RTGCPTR)SymInfo.Value + pThis->AddrKernelBase.FlatPtr), + aSymbols[i].pvVar, aSymbols[i].cbGuest); + if (RT_SUCCESS(rc)) + continue; + LogRel(("dbgDiggerLinuxIDmsg_QueryKernelLog: Reading '%s' at %RGv: %Rrc\n", aSymbols[i].pszSymbol, Addr.FlatPtr, rc)); + } + else + LogRel(("dbgDiggerLinuxIDmsg_QueryKernelLog: Error looking up '%s': %Rrc\n", aSymbols[i].pszSymbol, rc)); + rc = VERR_NOT_FOUND; + break; + } + + /* + * Some kernels don't expose the variables in kallsyms so we have to try disassemble + * some public helpers to get at the addresses. + * + * @todo: Maybe cache those values so we don't have to do the heavy work every time? + */ + if (rc == VERR_NOT_FOUND) + { + idxFirst = 0; + idxNext = 0; + rc = dbgDiggerLinuxQueryLogBufferPtrs(pThis, pUVM, pVMM, hMod, &GCPtrLogBuf, &cbLogBuf); + if (RT_FAILURE(rc)) + { + /* + * Last resort, scan for a known value which should appear only once in the kernel log buffer + * and try to deduce the boundaries from there. + */ + return dbgDiggerLinuxKrnlLogBufFindByNeedle(pThis, pUVM, pVMM, &GCPtrLogBuf, &cbLogBuf); + } + } + + return dbgDiggerLinuxKrnLogBufferProcess(pThis, pUVM, pVMM, GCPtrLogBuf, cbLogBuf, idxFirst, idxNext, + fFlags, cMessages, pszBuf, cbBuf, pcbActual); +} + +/** + * @interface_method_impl{DBGFOSIDMESG,pfnQueryKernelLog} + */ +static DECLCALLBACK(int) dbgDiggerLinuxIDmsg_QueryKernelLog(PDBGFOSIDMESG pThis, PUVM pUVM, PCVMMR3VTABLE pVMM, uint32_t fFlags, + uint32_t cMessages, char *pszBuf, size_t cbBuf, size_t *pcbActual) +{ + PDBGDIGGERLINUX pData = RT_FROM_MEMBER(pThis, DBGDIGGERLINUX, IDmesg); + + if (cMessages < 1) + return VERR_INVALID_PARAMETER; + + /* + * Resolve the symbols we need and read their values. + */ + RTDBGAS hAs = pVMM->pfnDBGFR3AsResolveAndRetain(pUVM, DBGF_AS_KERNEL); + RTDBGMOD hMod; + int rc = RTDbgAsModuleByName(hAs, "vmlinux", 0, &hMod); + RTDbgAsRelease(hAs); + + size_t cbActual = 0; + if (RT_SUCCESS(rc)) + { + /* + * Check whether the kernel log buffer is a simple char buffer or the newer + * record based implementation. + * The record based implementation was presumably introduced with kernel 3.4, + * see: http://thread.gmane.org/gmane.linux.kernel/1284184 + */ + if (dbgDiggerLinuxLogBufferIsAsciiBuffer(pData, pUVM, pVMM)) + rc = dbgDiggerLinuxLogBufferQueryAscii(pData, pUVM, pVMM, hMod, fFlags, cMessages, pszBuf, cbBuf, &cbActual); + else + rc = dbgDiggerLinuxLogBufferQueryRecords(pData, pUVM, pVMM, hMod, fFlags, cMessages, pszBuf, cbBuf, &cbActual); + + /* Release the module in any case. */ + RTDbgModRelease(hMod); + } + else + { + /* + * For the record based kernel versions we have a last resort heuristic which doesn't + * require any symbols, try that here. + */ + if (!dbgDiggerLinuxLogBufferIsAsciiBuffer(pData, pUVM, pVMM)) + { + RTGCPTR GCPtrLogBuf = 0; + uint32_t cbLogBuf = 0; + + rc = dbgDiggerLinuxKrnlLogBufFindByNeedle(pData, pUVM, pVMM, &GCPtrLogBuf, &cbLogBuf); + if (RT_SUCCESS(rc)) + rc = dbgDiggerLinuxKrnLogBufferProcess(pData, pUVM, pVMM, GCPtrLogBuf, cbLogBuf, 0 /*idxFirst*/, 0 /*idxNext*/, + fFlags, cMessages, pszBuf, cbBuf, &cbActual); + } + else + rc = VERR_NOT_FOUND; + } + + if (RT_FAILURE(rc) && rc != VERR_BUFFER_OVERFLOW) + return rc; + + if (pcbActual) + *pcbActual = cbActual; + + /* + * All VBox strings are UTF-8 and bad things may in theory happen if we + * pass bad UTF-8 to code which assumes it's all valid. So, we enforce + * UTF-8 upon the guest kernel messages here even if they (probably) have + * no defined code set in reality. + */ + if ( RT_SUCCESS(rc) + && cbActual <= cbBuf) + { + pszBuf[cbActual - 1] = '\0'; + RTStrPurgeEncoding(pszBuf); + return VINF_SUCCESS; + } + + if (cbBuf) + { + pszBuf[cbBuf - 1] = '\0'; + RTStrPurgeEncoding(pszBuf); + } + return VERR_BUFFER_OVERFLOW; +} + + +/** + * Worker destroying the config database. + */ +static DECLCALLBACK(int) dbgDiggerLinuxCfgDbDestroyWorker(PRTSTRSPACECORE pStr, void *pvUser) +{ + PDBGDIGGERLINUXCFGITEM pCfgItem = (PDBGDIGGERLINUXCFGITEM)pStr; + RTStrFree((char *)pCfgItem->Core.pszString); + RTMemFree(pCfgItem); + NOREF(pvUser); + return 0; +} + + +/** + * Destroy the config database. + * + * @param pThis The Linux digger data. + */ +static void dbgDiggerLinuxCfgDbDestroy(PDBGDIGGERLINUX pThis) +{ + RTStrSpaceDestroy(&pThis->hCfgDb, dbgDiggerLinuxCfgDbDestroyWorker, NULL); +} + + +/** + * @copydoc DBGFOSREG::pfnStackUnwindAssist + */ +static DECLCALLBACK(int) dbgDiggerLinuxStackUnwindAssist(PUVM pUVM, PCVMMR3VTABLE pVMM, void *pvData, VMCPUID idCpu, + PDBGFSTACKFRAME pFrame, PRTDBGUNWINDSTATE pState, PCCPUMCTX pInitialCtx, + RTDBGAS hAs, uint64_t *puScratch) +{ + RT_NOREF(pUVM, pVMM, pvData, idCpu, pFrame, pState, pInitialCtx, hAs, puScratch); + return VINF_SUCCESS; +} + + +/** + * @copydoc DBGFOSREG::pfnQueryInterface + */ +static DECLCALLBACK(void *) dbgDiggerLinuxQueryInterface(PUVM pUVM, PCVMMR3VTABLE pVMM, void *pvData, DBGFOSINTERFACE enmIf) +{ + PDBGDIGGERLINUX pThis = (PDBGDIGGERLINUX)pvData; + RT_NOREF(pUVM, pVMM); + + switch (enmIf) + { + case DBGFOSINTERFACE_DMESG: + return &pThis->IDmesg; + + default: + return NULL; + } +} + + +/** + * @copydoc DBGFOSREG::pfnQueryVersion + */ +static DECLCALLBACK(int) dbgDiggerLinuxQueryVersion(PUVM pUVM, PCVMMR3VTABLE pVMM, void *pvData, + char *pszVersion, size_t cchVersion) +{ + PDBGDIGGERLINUX pThis = (PDBGDIGGERLINUX)pvData; + Assert(pThis->fValid); + + /* + * It's all in the linux banner. + */ + int rc = pVMM->pfnDBGFR3MemReadString(pUVM, 0, &pThis->AddrLinuxBanner, pszVersion, cchVersion); + if (RT_SUCCESS(rc)) + { + char *pszEnd = RTStrEnd(pszVersion, cchVersion); + AssertReturn(pszEnd, VERR_BUFFER_OVERFLOW); + while ( pszEnd > pszVersion + && RT_C_IS_SPACE(pszEnd[-1])) + pszEnd--; + *pszEnd = '\0'; + } + else + RTStrPrintf(pszVersion, cchVersion, "DBGFR3MemRead -> %Rrc", rc); + + return rc; +} + + +/** + * @copydoc DBGFOSREG::pfnTerm + */ +static DECLCALLBACK(void) dbgDiggerLinuxTerm(PUVM pUVM, PCVMMR3VTABLE pVMM, void *pvData) +{ + PDBGDIGGERLINUX pThis = (PDBGDIGGERLINUX)pvData; + Assert(pThis->fValid); + + /* + * Destroy configuration database. + */ + dbgDiggerLinuxCfgDbDestroy(pThis); + + /* + * Unlink and release our modules. + */ + RTDBGAS hDbgAs = pVMM->pfnDBGFR3AsResolveAndRetain(pUVM, DBGF_AS_KERNEL); + if (hDbgAs != NIL_RTDBGAS) + { + uint32_t iMod = RTDbgAsModuleCount(hDbgAs); + while (iMod-- > 0) + { + RTDBGMOD hMod = RTDbgAsModuleByIndex(hDbgAs, iMod); + if (hMod != NIL_RTDBGMOD) + { + if (RTDbgModGetTag(hMod) == DIG_LNX_MOD_TAG) + { + int rc = RTDbgAsModuleUnlink(hDbgAs, hMod); + AssertRC(rc); + } + RTDbgModRelease(hMod); + } + } + RTDbgAsRelease(hDbgAs); + } + + pThis->fValid = false; +} + + +/** + * @copydoc DBGFOSREG::pfnRefresh + */ +static DECLCALLBACK(int) dbgDiggerLinuxRefresh(PUVM pUVM, PCVMMR3VTABLE pVMM, void *pvData) +{ + PDBGDIGGERLINUX pThis = (PDBGDIGGERLINUX)pvData; + RT_NOREF(pThis); + Assert(pThis->fValid); + + /* + * For now we'll flush and reload everything. + */ + dbgDiggerLinuxTerm(pUVM, pVMM, pvData); + return dbgDiggerLinuxInit(pUVM, pVMM, pvData); +} + + +/** + * Worker for dbgDiggerLinuxFindStartOfNamesAndSymbolCount that update the + * digger data. + * + * @returns VINF_SUCCESS. + * @param pThis The Linux digger data to update. + * @param pVMM The VMM function table. + * @param pAddrKernelNames The kallsyms_names address. + * @param cKernelSymbols The number of kernel symbol. + * @param cbAddress The guest address size. + */ +static int dbgDiggerLinuxFoundStartOfNames(PDBGDIGGERLINUX pThis, PCVMMR3VTABLE pVMM, PCDBGFADDRESS pAddrKernelNames, + uint32_t cKernelSymbols, uint32_t cbAddress) +{ + pThis->cKernelSymbols = cKernelSymbols; + pThis->AddrKernelNames = *pAddrKernelNames; + pThis->AddrKernelAddresses = *pAddrKernelNames; + uint32_t cbSymbolsSkip = (pThis->fRelKrnlAddr ? 2 : 1) * cbAddress; /* Relative addressing introduces kallsyms_relative_base. */ + uint32_t cbOffsets = pThis->fRelKrnlAddr ? sizeof(int32_t) : cbAddress; /* Offsets are always 32bits wide for relative addressing. */ + uint32_t cbAlign = 0; + + /* + * If the number of symbols is odd there is padding to align the following guest pointer + * sized data properly on 64bit systems with relative addressing. + */ + if ( pThis->fRelKrnlAddr + && pThis->f64Bit + && (pThis->cKernelSymbols & 1)) + cbAlign = sizeof(int32_t); + pVMM->pfnDBGFR3AddrSub(&pThis->AddrKernelAddresses, cKernelSymbols * cbOffsets + cbSymbolsSkip + cbAlign); + + Log(("dbgDiggerLinuxFoundStartOfNames: AddrKernelAddresses=%RGv\n" + "dbgDiggerLinuxFoundStartOfNames: cKernelSymbols=%#x (at %RGv)\n" + "dbgDiggerLinuxFoundStartOfNames: AddrKernelName=%RGv\n", + pThis->AddrKernelAddresses.FlatPtr, + pThis->cKernelSymbols, pThis->AddrKernelNames.FlatPtr - cbAddress, + pThis->AddrKernelNames.FlatPtr)); + return VINF_SUCCESS; +} + + +/** + * Tries to find the address of the kallsyms_names, kallsyms_num_syms and + * kallsyms_addresses symbols. + * + * The kallsyms_num_syms is read and stored in pThis->cKernelSymbols, while the + * addresses of the other two are stored as pThis->AddrKernelNames and + * pThis->AddrKernelAddresses. + * + * @returns VBox status code, success indicating that all three variables have + * been found and taken down. + * @param pUVM The user mode VM handle. + * @param pVMM The VMM function table. + * @param pThis The Linux digger data. + * @param pHitAddr An address we think is inside kallsyms_names. + */ +static int dbgDiggerLinuxFindStartOfNamesAndSymbolCount(PUVM pUVM, PCVMMR3VTABLE pVMM, PDBGDIGGERLINUX pThis, + PCDBGFADDRESS pHitAddr) +{ + /* + * Search backwards in chunks. + */ + union + { + uint8_t ab[0x1000]; + uint32_t au32[0x1000 / sizeof(uint32_t)]; + uint64_t au64[0x1000 / sizeof(uint64_t)]; + } uBuf; + uint32_t cbLeft = LNX_MAX_KALLSYMS_NAMES_SIZE; + uint32_t cbBuf = pHitAddr->FlatPtr & (sizeof(uBuf) - 1); + DBGFADDRESS CurAddr = *pHitAddr; + pVMM->pfnDBGFR3AddrSub(&CurAddr, cbBuf); + cbBuf += sizeof(uint64_t) - 1; /* In case our kobj hit is in the first 4/8 bytes. */ + for (;;) + { + int rc = pVMM->pfnDBGFR3MemRead(pUVM, 0 /*idCpu*/, &CurAddr, &uBuf, sizeof(uBuf)); + if (RT_FAILURE(rc)) + return rc; + + /* + * Since Linux 4.6 there are two different methods to store the kallsyms addresses + * in the image. + * + * The first and longer existing method is to store the absolute addresses in an + * array starting at kallsyms_addresses followed by a field which stores the number + * of kernel symbols called kallsyms_num_syms. + * The newer method is to use offsets stored in kallsyms_offsets and have a base pointer + * to relate the offsets to called kallsyms_relative_base. One entry in kallsyms_offsets is + * always 32bit wide regardless of the guest pointer size (this halves the table on 64bit + * systems) but means more work for us for the 64bit case. + * + * When absolute addresses are used the following assumptions hold: + * + * We assume that the three symbols are aligned on guest pointer boundary. + * + * The boundary between the two tables should be noticable as the number + * is unlikely to be more than 16 millions, there will be at least one zero + * byte where it is, 64-bit will have 5 zero bytes. Zero bytes aren't all + * that common in the kallsyms_names table. + * + * Also the kallsyms_names table starts with a length byte, which means + * we're likely to see a byte in the range 1..31. + * + * The kallsyms_addresses are mostly sorted (except for the start where the + * absolute symbols are), so we'll spot a bunch of kernel addresses + * immediately preceeding the kallsyms_num_syms field. + * + * Lazy bird: If kallsyms_num_syms is on a buffer boundrary, we skip + * the check for kernel addresses preceeding it. + * + * For relative offsets most of the assumptions from above are true too + * except that we have to distinguish between the relative base address and the offsets. + * Every observed kernel has a valid kernel address fo the relative base and kallsyms_relative_base + * always comes before kallsyms_num_syms and is aligned on a guest pointer boundary. + * Offsets are stored before kallsyms_relative_base and don't contain valid kernel addresses. + * + * To distinguish between absolute and relative offsetting we check the data before a candidate + * for kallsyms_num_syms. If all entries before the kallsyms_num_syms candidate are valid kernel + * addresses absolute addresses are assumed. If this is not the case but the first entry before + * kallsyms_num_syms is a valid kernel address we check whether the data before and the possible + * relative base form a valid kernel address and assume relative offsets. + * + * Other notable changes between various Linux kernel versions: + * + * 4.20.0+: Commit 80ffbaa5b1bd98e80e3239a3b8cfda2da433009a made kallsyms_num_syms 32bit + * even on 64bit systems but the alignment of the variables makes the code below work for now + * (tested with a 5.4 and 5.12 kernel) do we keep it that way to avoid making the code even + * messy. + */ + if (pThis->f64Bit) + { + uint32_t i = cbBuf / sizeof(uint64_t) - 1; + while (i-- > 0) + if ( uBuf.au64[i] <= LNX_MAX_KALLSYMS_SYMBOLS + && uBuf.au64[i] >= LNX_MIN_KALLSYMS_SYMBOLS) + { + uint8_t *pb = (uint8_t *)&uBuf.au64[i + 1]; + if ( pb[0] <= LNX_MAX_KALLSYMS_ENC_LENGTH + && pb[0] >= LNX_MIN_KALLSYMS_ENC_LENGTH) + { + /* + * Check whether we have a valid kernel address and try to distinguish + * whether the kernel uses relative offsetting or absolute addresses. + */ + if ( (i >= 1 && LNX64_VALID_ADDRESS(uBuf.au64[i - 1])) + && (i >= 2 && !LNX64_VALID_ADDRESS(uBuf.au64[i - 2])) + && (i >= 3 && !LNX64_VALID_ADDRESS(uBuf.au64[i - 3]))) + { + RTGCUINTPTR uKrnlRelBase = uBuf.au64[i - 1]; + DBGFADDRESS RelAddr = CurAddr; + int32_t aiRelOff[3]; + rc = pVMM->pfnDBGFR3MemRead(pUVM, 0 /*idCpu*/, + pVMM->pfnDBGFR3AddrAdd(&RelAddr, + (i - 1) * sizeof(uint64_t) - sizeof(aiRelOff)), + &aiRelOff[0], sizeof(aiRelOff)); + if ( RT_SUCCESS(rc) + && LNX64_VALID_ADDRESS(uKrnlRelBase + aiRelOff[0]) + && LNX64_VALID_ADDRESS(uKrnlRelBase + aiRelOff[1]) + && LNX64_VALID_ADDRESS(uKrnlRelBase + aiRelOff[2])) + { + Log(("dbgDiggerLinuxFindStartOfNamesAndSymbolCount: relative base %RGv (at %RGv)\n", + uKrnlRelBase, CurAddr.FlatPtr + (i - 1) * sizeof(uint64_t))); + pThis->fRelKrnlAddr = true; + pThis->uKernelRelativeBase = uKrnlRelBase; + return dbgDiggerLinuxFoundStartOfNames(pThis, pVMM, + pVMM->pfnDBGFR3AddrAdd(&CurAddr, (i + 1) * sizeof(uint64_t)), + (uint32_t)uBuf.au64[i], sizeof(uint64_t)); + } + } + + if ( (i <= 0 || LNX64_VALID_ADDRESS(uBuf.au64[i - 1])) + && (i <= 1 || LNX64_VALID_ADDRESS(uBuf.au64[i - 2])) + && (i <= 2 || LNX64_VALID_ADDRESS(uBuf.au64[i - 3]))) + return dbgDiggerLinuxFoundStartOfNames(pThis, pVMM, + pVMM->pfnDBGFR3AddrAdd(&CurAddr, (i + 1) * sizeof(uint64_t)), + (uint32_t)uBuf.au64[i], sizeof(uint64_t)); + } + } + } + else + { + uint32_t i = cbBuf / sizeof(uint32_t) - 1; + while (i-- > 0) + if ( uBuf.au32[i] <= LNX_MAX_KALLSYMS_SYMBOLS + && uBuf.au32[i] >= LNX_MIN_KALLSYMS_SYMBOLS) + { + uint8_t *pb = (uint8_t *)&uBuf.au32[i + 1]; + if ( pb[0] <= LNX_MAX_KALLSYMS_ENC_LENGTH + && pb[0] >= LNX_MIN_KALLSYMS_ENC_LENGTH) + { + /* Check for relative base addressing. */ + if (i >= 1 && LNX32_VALID_ADDRESS(uBuf.au32[i - 1])) + { + RTGCUINTPTR uKrnlRelBase = uBuf.au32[i - 1]; + if ( (i <= 1 || LNX32_VALID_ADDRESS(uKrnlRelBase + uBuf.au32[i - 2])) + && (i <= 2 || LNX32_VALID_ADDRESS(uKrnlRelBase + uBuf.au32[i - 3]))) + { + Log(("dbgDiggerLinuxFindStartOfNamesAndSymbolCount: relative base %RGv (at %RGv)\n", + uKrnlRelBase, CurAddr.FlatPtr + (i - 1) * sizeof(uint32_t))); + pThis->fRelKrnlAddr = true; + pThis->uKernelRelativeBase = uKrnlRelBase; + return dbgDiggerLinuxFoundStartOfNames(pThis, pVMM, + pVMM->pfnDBGFR3AddrAdd(&CurAddr, (i + 1) * sizeof(uint32_t)), + uBuf.au32[i], sizeof(uint32_t)); + } + } + + if ( (i <= 0 || LNX32_VALID_ADDRESS(uBuf.au32[i - 1])) + && (i <= 1 || LNX32_VALID_ADDRESS(uBuf.au32[i - 2])) + && (i <= 2 || LNX32_VALID_ADDRESS(uBuf.au32[i - 3]))) + return dbgDiggerLinuxFoundStartOfNames(pThis, pVMM, + pVMM->pfnDBGFR3AddrAdd(&CurAddr, (i + 1) * sizeof(uint32_t)), + uBuf.au32[i], sizeof(uint32_t)); + } + } + } + + /* + * Advance + */ + if (RT_UNLIKELY(cbLeft <= sizeof(uBuf))) + { + Log(("dbgDiggerLinuxFindStartOfNamesAndSymbolCount: failed (pHitAddr=%RGv)\n", pHitAddr->FlatPtr)); + return VERR_NOT_FOUND; + } + cbLeft -= sizeof(uBuf); + pVMM->pfnDBGFR3AddrSub(&CurAddr, sizeof(uBuf)); + cbBuf = sizeof(uBuf); + } +} + + +/** + * Worker for dbgDiggerLinuxFindEndNames that records the findings. + * + * @returns VINF_SUCCESS + * @param pThis The linux digger data to update. + * @param pVMM The VMM function table. + * @param pAddrMarkers The address of the marker (kallsyms_markers). + * @param cbMarkerEntry The size of a marker entry (32-bit or 64-bit). + */ +static int dbgDiggerLinuxFoundMarkers(PDBGDIGGERLINUX pThis, PCVMMR3VTABLE pVMM, + PCDBGFADDRESS pAddrMarkers, uint32_t cbMarkerEntry) +{ + pThis->cbKernelNames = pAddrMarkers->FlatPtr - pThis->AddrKernelNames.FlatPtr; + pThis->AddrKernelNameMarkers = *pAddrMarkers; + pThis->cKernelNameMarkers = RT_ALIGN_32(pThis->cKernelSymbols, 256) / 256; + pThis->AddrKernelTokenTable = *pAddrMarkers; + pVMM->pfnDBGFR3AddrAdd(&pThis->AddrKernelTokenTable, pThis->cKernelNameMarkers * cbMarkerEntry); + + Log(("dbgDiggerLinuxFoundMarkers: AddrKernelNames=%RGv cbKernelNames=%#x\n" + "dbgDiggerLinuxFoundMarkers: AddrKernelNameMarkers=%RGv cKernelNameMarkers=%#x\n" + "dbgDiggerLinuxFoundMarkers: AddrKernelTokenTable=%RGv\n", + pThis->AddrKernelNames.FlatPtr, pThis->cbKernelNames, + pThis->AddrKernelNameMarkers.FlatPtr, pThis->cKernelNameMarkers, + pThis->AddrKernelTokenTable.FlatPtr)); + return VINF_SUCCESS; +} + + +/** + * Tries to find the end of kallsyms_names and thereby the start of + * kallsyms_markers and kallsyms_token_table. + * + * The kallsyms_names size is stored in pThis->cbKernelNames, the addresses of + * the two other symbols in pThis->AddrKernelNameMarkers and + * pThis->AddrKernelTokenTable. The number of marker entries is stored in + * pThis->cKernelNameMarkers. + * + * @returns VBox status code, success indicating that all three variables have + * been found and taken down. + * @param pUVM The user mode VM handle. + * @param pVMM The VMM function table. + * @param pThis The Linux digger data. + * @param pHitAddr An address we think is inside kallsyms_names. + */ +static int dbgDiggerLinuxFindEndOfNamesAndMore(PUVM pUVM, PCVMMR3VTABLE pVMM, PDBGDIGGERLINUX pThis, PCDBGFADDRESS pHitAddr) +{ + /* + * Search forward in chunks. + */ + union + { + uint8_t ab[0x1000]; + uint32_t au32[0x1000 / sizeof(uint32_t)]; + uint64_t au64[0x1000 / sizeof(uint64_t)]; + } uBuf; + bool fPendingZeroHit = false; + uint32_t cbLeft = LNX_MAX_KALLSYMS_NAMES_SIZE + sizeof(uBuf); + uint32_t offBuf = pHitAddr->FlatPtr & (sizeof(uBuf) - 1); + DBGFADDRESS CurAddr = *pHitAddr; + pVMM->pfnDBGFR3AddrSub(&CurAddr, offBuf); + for (;;) + { + int rc = pVMM->pfnDBGFR3MemRead(pUVM, 0 /*idCpu*/, &CurAddr, &uBuf, sizeof(uBuf)); + if (RT_FAILURE(rc)) + return rc; + + /* + * The kallsyms_names table is followed by kallsyms_markers we assume, + * using sizeof(unsigned long) alignment like the preceeding symbols. + * + * The kallsyms_markers table has entried sizeof(unsigned long) and + * contains offsets into kallsyms_names. The kallsyms_markers used to + * index kallsyms_names and reduce seek time when looking up the name + * of an address/symbol. Each entry in kallsyms_markers covers 256 + * symbol names. + * + * Because of this, the first entry is always zero and all the entries + * are ascending. It also follows that the size of the table can be + * calculated from kallsyms_num_syms. + * + * Note! We could also have walked kallsyms_names by skipping + * kallsyms_num_syms names, but this is faster and we will + * validate the encoded names later. + * + * git commit 80ffbaa5b1bd98e80e3239a3b8cfda2da433009a (which became 4.20+) makes kallsyms_markers + * and kallsyms_num_syms uint32_t, even on 64bit systems. Take that into account. + */ + if ( pThis->f64Bit + && pThis->uKrnlVer < LNX_MK_VER(4, 20, 0)) + { + if ( RT_UNLIKELY(fPendingZeroHit) + && uBuf.au64[0] >= (LNX_MIN_KALLSYMS_ENC_LENGTH + 1) * 256 + && uBuf.au64[0] <= (LNX_MAX_KALLSYMS_ENC_LENGTH + 1) * 256) + return dbgDiggerLinuxFoundMarkers(pThis, pVMM, + pVMM->pfnDBGFR3AddrSub(&CurAddr, sizeof(uint64_t)), sizeof(uint64_t)); + + uint32_t const cEntries = sizeof(uBuf) / sizeof(uint64_t); + for (uint32_t i = offBuf / sizeof(uint64_t); i < cEntries; i++) + if (uBuf.au64[i] == 0) + { + if (RT_UNLIKELY(i + 1 >= cEntries)) + { + fPendingZeroHit = true; + break; + } + if ( uBuf.au64[i + 1] >= (LNX_MIN_KALLSYMS_ENC_LENGTH + 1) * 256 + && uBuf.au64[i + 1] <= (LNX_MAX_KALLSYMS_ENC_LENGTH + 1) * 256) + return dbgDiggerLinuxFoundMarkers(pThis, pVMM, + pVMM->pfnDBGFR3AddrAdd(&CurAddr, i * sizeof(uint64_t)), sizeof(uint64_t)); + } + } + else + { + if ( RT_UNLIKELY(fPendingZeroHit) + && uBuf.au32[0] >= (LNX_MIN_KALLSYMS_ENC_LENGTH + 1) * 256 + && uBuf.au32[0] <= (LNX_MAX_KALLSYMS_ENC_LENGTH + 1) * 256) + return dbgDiggerLinuxFoundMarkers(pThis, pVMM, + pVMM->pfnDBGFR3AddrSub(&CurAddr, sizeof(uint32_t)), sizeof(uint32_t)); + + uint32_t const cEntries = sizeof(uBuf) / sizeof(uint32_t); + for (uint32_t i = offBuf / sizeof(uint32_t); i < cEntries; i++) + if (uBuf.au32[i] == 0) + { + if (RT_UNLIKELY(i + 1 >= cEntries)) + { + fPendingZeroHit = true; + break; + } + if ( uBuf.au32[i + 1] >= (LNX_MIN_KALLSYMS_ENC_LENGTH + 1) * 256 + && uBuf.au32[i + 1] <= (LNX_MAX_KALLSYMS_ENC_LENGTH + 1) * 256) + return dbgDiggerLinuxFoundMarkers(pThis, pVMM, + pVMM->pfnDBGFR3AddrAdd(&CurAddr, i * sizeof(uint32_t)), sizeof(uint32_t)); + } + } + + /* + * Advance + */ + if (RT_UNLIKELY(cbLeft <= sizeof(uBuf))) + { + Log(("dbgDiggerLinuxFindEndOfNamesAndMore: failed (pHitAddr=%RGv)\n", pHitAddr->FlatPtr)); + return VERR_NOT_FOUND; + } + cbLeft -= sizeof(uBuf); + pVMM->pfnDBGFR3AddrAdd(&CurAddr, sizeof(uBuf)); + offBuf = 0; + } +} + + +/** + * Locates the kallsyms_token_index table. + * + * Storing the address in pThis->AddrKernelTokenIndex and the size of the token + * table in pThis->cbKernelTokenTable. + * + * @returns VBox status code. + * @param pUVM The user mode VM handle. + * @param pVMM The VMM function table. + * @param pThis The Linux digger data. + */ +static int dbgDiggerLinuxFindTokenIndex(PUVM pUVM, PCVMMR3VTABLE pVMM, PDBGDIGGERLINUX pThis) +{ + /* + * The kallsyms_token_table is very much like a string table. Due to the + * nature of the compression algorithm it is reasonably short (one example + * here is 853 bytes), so we'll not be reading it in chunks but in full. + * To be on the safe side, we read 8KB, ASSUMING we won't run into unmapped + * memory or any other nasty stuff... + */ + union + { + uint8_t ab[0x2000]; + uint16_t au16[0x2000 / sizeof(uint16_t)]; + } uBuf; + DBGFADDRESS CurAddr = pThis->AddrKernelTokenTable; + int rc = pVMM->pfnDBGFR3MemRead(pUVM, 0 /*idCpu*/, &CurAddr, &uBuf, sizeof(uBuf)); + if (RT_FAILURE(rc)) + return rc; + + /* + * We've got two choices here, either walk the string table or look for + * the next structure, kallsyms_token_index. + * + * The token index is a table of 256 uint16_t entries (index by bytes + * from kallsyms_names) that gives offsets in kallsyms_token_table. It + * starts with a zero entry and the following entries are sorted in + * ascending order. The range of the entries are reasonably small since + * kallsyms_token_table is small. + * + * The alignment seems to be sizeof(unsigned long), just like + * kallsyms_token_table. + * + * So, we start by looking for a zero 16-bit entry. + */ + uint32_t cIncr = (pThis->f64Bit ? sizeof(uint64_t) : sizeof(uint32_t)) / sizeof(uint16_t); + + for (uint32_t i = 0; i < sizeof(uBuf) / sizeof(uint16_t) - 16; i += cIncr) + if ( uBuf.au16[i] == 0 + && uBuf.au16[i + 1] > 0 + && uBuf.au16[i + 1] <= LNX_MAX_KALLSYMS_TOKEN_LEN + && (uint16_t)(uBuf.au16[i + 2] - uBuf.au16[i + 1] - 1U) <= (uint16_t)LNX_MAX_KALLSYMS_TOKEN_LEN + && (uint16_t)(uBuf.au16[i + 3] - uBuf.au16[i + 2] - 1U) <= (uint16_t)LNX_MAX_KALLSYMS_TOKEN_LEN + && (uint16_t)(uBuf.au16[i + 4] - uBuf.au16[i + 3] - 1U) <= (uint16_t)LNX_MAX_KALLSYMS_TOKEN_LEN + && (uint16_t)(uBuf.au16[i + 5] - uBuf.au16[i + 4] - 1U) <= (uint16_t)LNX_MAX_KALLSYMS_TOKEN_LEN + && (uint16_t)(uBuf.au16[i + 6] - uBuf.au16[i + 5] - 1U) <= (uint16_t)LNX_MAX_KALLSYMS_TOKEN_LEN + ) + { + pThis->AddrKernelTokenIndex = CurAddr; + pVMM->pfnDBGFR3AddrAdd(&pThis->AddrKernelTokenIndex, i * sizeof(uint16_t)); + pThis->cbKernelTokenTable = i * sizeof(uint16_t); + return VINF_SUCCESS; + } + + Log(("dbgDiggerLinuxFindTokenIndex: Failed (%RGv..%RGv)\n", CurAddr.FlatPtr, CurAddr.FlatPtr + (RTGCUINTPTR)sizeof(uBuf))); + return VERR_NOT_FOUND; +} + + +/** + * Loads the kernel symbols from the given kallsyms offset table decoding the symbol names + * (worker common for dbgDiggerLinuxLoadKernelSymbolsAbsolute() and dbgDiggerLinuxLoadKernelSymbolsRelative()). + * + * @returns VBox status code. + * @param pUVM The user mode VM handle. + * @param pVMM The VMM function table. + * @param pThis The Linux digger data. + * @param uKernelStart Flat kernel start address. + * @param cbKernel Size of the kernel in bytes. + * @param pauSymOff Pointer to the array of symbol offsets in the kallsyms table + * relative to the start of the kernel. + */ +static int dbgDiggerLinuxLoadKernelSymbolsWorker(PUVM pUVM, PCVMMR3VTABLE pVMM, PDBGDIGGERLINUX pThis, RTGCUINTPTR uKernelStart, + RTGCUINTPTR cbKernel, RTGCUINTPTR *pauSymOff) +{ + uint8_t *pbNames = (uint8_t *)RTMemAllocZ(pThis->cbKernelNames); + int rc = pVMM->pfnDBGFR3MemRead(pUVM, 0 /*idCpu*/, &pThis->AddrKernelNames, pbNames, pThis->cbKernelNames); + if (RT_SUCCESS(rc)) + { + char *pszzTokens = (char *)RTMemAllocZ(pThis->cbKernelTokenTable); + rc = pVMM->pfnDBGFR3MemRead(pUVM, 0 /*idCpu*/, &pThis->AddrKernelTokenTable, pszzTokens, pThis->cbKernelTokenTable); + if (RT_SUCCESS(rc)) + { + uint16_t *paoffTokens = (uint16_t *)RTMemAllocZ(256 * sizeof(uint16_t)); + rc = pVMM->pfnDBGFR3MemRead(pUVM, 0 /*idCpu*/, &pThis->AddrKernelTokenIndex, paoffTokens, 256 * sizeof(uint16_t)); + if (RT_SUCCESS(rc)) + { + /* + * Create a module for the kernel. + */ + RTDBGMOD hMod; + rc = RTDbgModCreate(&hMod, "vmlinux", cbKernel, 0 /*fFlags*/); + if (RT_SUCCESS(rc)) + { + rc = RTDbgModSetTag(hMod, DIG_LNX_MOD_TAG); AssertRC(rc); + rc = VINF_SUCCESS; + + /* + * Enumerate the symbols. + */ + uint32_t offName = 0; + uint32_t cLeft = pThis->cKernelSymbols; + while (cLeft-- > 0 && RT_SUCCESS(rc)) + { + /* Decode the symbol name first. */ + if (RT_LIKELY(offName < pThis->cbKernelNames)) + { + uint8_t cbName = pbNames[offName++]; + if (RT_LIKELY(offName + cbName <= pThis->cbKernelNames)) + { + char szSymbol[4096]; + uint32_t offSymbol = 0; + while (cbName-- > 0) + { + uint8_t bEnc = pbNames[offName++]; + uint16_t offToken = paoffTokens[bEnc]; + if (RT_LIKELY(offToken < pThis->cbKernelTokenTable)) + { + const char *pszToken = &pszzTokens[offToken]; + char ch; + while ((ch = *pszToken++) != '\0') + if (offSymbol < sizeof(szSymbol) - 1) + szSymbol[offSymbol++] = ch; + } + else + { + rc = VERR_INVALID_UTF8_ENCODING; + break; + } + } + szSymbol[offSymbol < sizeof(szSymbol) ? offSymbol : sizeof(szSymbol) - 1] = '\0'; + + /* The offset. */ + RTGCUINTPTR uSymOff = *pauSymOff; + pauSymOff++; + + /* Add it without the type char. */ + if (uSymOff <= cbKernel) + { + rc = RTDbgModSymbolAdd(hMod, &szSymbol[1], RTDBGSEGIDX_RVA, uSymOff, + 0 /*cb*/, 0 /*fFlags*/, NULL); + if (RT_FAILURE(rc)) + { + if ( rc == VERR_DBG_SYMBOL_NAME_OUT_OF_RANGE + || rc == VERR_DBG_INVALID_RVA + || rc == VERR_DBG_ADDRESS_CONFLICT + || rc == VERR_DBG_DUPLICATE_SYMBOL) + { + Log2(("dbgDiggerLinuxLoadKernelSymbols: RTDbgModSymbolAdd(,%s,) failed %Rrc (ignored)\n", szSymbol, rc)); + rc = VINF_SUCCESS; + } + else + Log(("dbgDiggerLinuxLoadKernelSymbols: RTDbgModSymbolAdd(,%s,) failed %Rrc\n", szSymbol, rc)); + } + } + } + else + { + rc = VERR_END_OF_STRING; + Log(("dbgDiggerLinuxLoadKernelSymbols: offName=%#x cLeft=%#x cbName=%#x cbKernelNames=%#x\n", + offName, cLeft, cbName, pThis->cbKernelNames)); + } + } + else + { + rc = VERR_END_OF_STRING; + Log(("dbgDiggerLinuxLoadKernelSymbols: offName=%#x cLeft=%#x cbKernelNames=%#x\n", + offName, cLeft, pThis->cbKernelNames)); + } + } + + /* + * Link the module into the address space. + */ + if (RT_SUCCESS(rc)) + { + RTDBGAS hAs = pVMM->pfnDBGFR3AsResolveAndRetain(pUVM, DBGF_AS_KERNEL); + if (hAs != NIL_RTDBGAS) + rc = RTDbgAsModuleLink(hAs, hMod, uKernelStart, RTDBGASLINK_FLAGS_REPLACE); + else + rc = VERR_INTERNAL_ERROR; + RTDbgAsRelease(hAs); + } + else + Log(("dbgDiggerLinuxLoadKernelSymbols: Failed: %Rrc\n", rc)); + RTDbgModRelease(hMod); + } + else + Log(("dbgDiggerLinuxLoadKernelSymbols: RTDbgModCreate failed: %Rrc\n", rc)); + } + else + Log(("dbgDiggerLinuxLoadKernelSymbols: Reading token index at %RGv failed: %Rrc\n", + pThis->AddrKernelTokenIndex.FlatPtr, rc)); + RTMemFree(paoffTokens); + } + else + Log(("dbgDiggerLinuxLoadKernelSymbols: Reading token table at %RGv failed: %Rrc\n", + pThis->AddrKernelTokenTable.FlatPtr, rc)); + RTMemFree(pszzTokens); + } + else + Log(("dbgDiggerLinuxLoadKernelSymbols: Reading encoded names at %RGv failed: %Rrc\n", + pThis->AddrKernelNames.FlatPtr, rc)); + RTMemFree(pbNames); + + return rc; +} + +/** + * Loads the kernel symbols from the kallsyms table if it contains absolute addresses + * + * @returns VBox status code. + * @param pUVM The user mode VM handle. + * @param pVMM The VMM function table. + * @param pThis The Linux digger data. + */ +static int dbgDiggerLinuxLoadKernelSymbolsAbsolute(PUVM pUVM, PCVMMR3VTABLE pVMM, PDBGDIGGERLINUX pThis) +{ + /* + * Allocate memory for temporary table copies, reading the tables as we go. + */ + uint32_t const cbGuestAddr = pThis->f64Bit ? sizeof(uint64_t) : sizeof(uint32_t); + void *pvAddresses = RTMemAllocZ(pThis->cKernelSymbols * cbGuestAddr); + int rc = pVMM->pfnDBGFR3MemRead(pUVM, 0 /*idCpu*/, &pThis->AddrKernelAddresses, + pvAddresses, pThis->cKernelSymbols * cbGuestAddr); + if (RT_SUCCESS(rc)) + { + /* + * Figure out the kernel start and end and convert the absolute addresses to relative offsets. + */ + RTGCUINTPTR uKernelStart = pThis->AddrKernelAddresses.FlatPtr; + RTGCUINTPTR uKernelEnd = pThis->AddrKernelTokenIndex.FlatPtr + 256 * sizeof(uint16_t); + RTGCUINTPTR *pauSymOff = (RTGCUINTPTR *)RTMemTmpAllocZ(pThis->cKernelSymbols * sizeof(RTGCUINTPTR)); + uint32_t i; + if (cbGuestAddr == sizeof(uint64_t)) + { + uint64_t *pauAddrs = (uint64_t *)pvAddresses; + for (i = 0; i < pThis->cKernelSymbols; i++) + if ( pauAddrs[i] < uKernelStart + && LNX64_VALID_ADDRESS(pauAddrs[i]) + && uKernelStart - pauAddrs[i] < LNX_MAX_KERNEL_SIZE) + uKernelStart = pauAddrs[i]; + + for (i = pThis->cKernelSymbols - 1; i > 0; i--) + if ( pauAddrs[i] > uKernelEnd + && LNX64_VALID_ADDRESS(pauAddrs[i]) + && pauAddrs[i] - uKernelEnd < LNX_MAX_KERNEL_SIZE) + uKernelEnd = pauAddrs[i]; + + for (i = 0; i < pThis->cKernelSymbols; i++) + pauSymOff[i] = pauAddrs[i] - uKernelStart; + } + else + { + uint32_t *pauAddrs = (uint32_t *)pvAddresses; + for (i = 0; i < pThis->cKernelSymbols; i++) + if ( pauAddrs[i] < uKernelStart + && LNX32_VALID_ADDRESS(pauAddrs[i]) + && uKernelStart - pauAddrs[i] < LNX_MAX_KERNEL_SIZE) + uKernelStart = pauAddrs[i]; + + for (i = pThis->cKernelSymbols - 1; i > 0; i--) + if ( pauAddrs[i] > uKernelEnd + && LNX32_VALID_ADDRESS(pauAddrs[i]) + && pauAddrs[i] - uKernelEnd < LNX_MAX_KERNEL_SIZE) + uKernelEnd = pauAddrs[i]; + + for (i = 0; i < pThis->cKernelSymbols; i++) + pauSymOff[i] = pauAddrs[i] - uKernelStart; + } + + RTGCUINTPTR cbKernel = uKernelEnd - uKernelStart; + pThis->cbKernel = (uint32_t)cbKernel; + pVMM->pfnDBGFR3AddrFromFlat(pUVM, &pThis->AddrKernelBase, uKernelStart); + Log(("dbgDiggerLinuxLoadKernelSymbolsAbsolute: uKernelStart=%RGv cbKernel=%#x\n", uKernelStart, cbKernel)); + + rc = dbgDiggerLinuxLoadKernelSymbolsWorker(pUVM, pVMM, pThis, uKernelStart, cbKernel, pauSymOff); + if (RT_FAILURE(rc)) + Log(("dbgDiggerLinuxLoadKernelSymbolsAbsolute: Loading symbols from given offset table failed: %Rrc\n", rc)); + RTMemTmpFree(pauSymOff); + } + else + Log(("dbgDiggerLinuxLoadKernelSymbolsAbsolute: Reading symbol addresses at %RGv failed: %Rrc\n", + pThis->AddrKernelAddresses.FlatPtr, rc)); + RTMemFree(pvAddresses); + + return rc; +} + + +/** + * Loads the kernel symbols from the kallsyms table if it contains absolute addresses + * + * @returns VBox status code. + * @param pUVM The user mode VM handle. + * @param pVMM The VMM function table. + * @param pThis The Linux digger data. + */ +static int dbgDiggerLinuxLoadKernelSymbolsRelative(PUVM pUVM, PCVMMR3VTABLE pVMM, PDBGDIGGERLINUX pThis) +{ + /* + * Allocate memory for temporary table copies, reading the tables as we go. + */ + int32_t *pai32Offsets = (int32_t *)RTMemAllocZ(pThis->cKernelSymbols * sizeof(int32_t)); + int rc = pVMM->pfnDBGFR3MemRead(pUVM, 0 /*idCpu*/, &pThis->AddrKernelAddresses, + pai32Offsets, pThis->cKernelSymbols * sizeof(int32_t)); + if (RT_SUCCESS(rc)) + { + /* + * Figure out the kernel start and end and convert the absolute addresses to relative offsets. + */ + RTGCUINTPTR uKernelStart = pThis->AddrKernelAddresses.FlatPtr; + RTGCUINTPTR uKernelEnd = pThis->AddrKernelTokenIndex.FlatPtr + 256 * sizeof(uint16_t); + RTGCUINTPTR *pauSymOff = (RTGCUINTPTR *)RTMemTmpAllocZ(pThis->cKernelSymbols * sizeof(RTGCUINTPTR)); + uint32_t i; + + for (i = 0; i < pThis->cKernelSymbols; i++) + { + RTGCUINTPTR uSymAddr = dbgDiggerLinuxConvOffsetToAddr(pThis, pai32Offsets[i]); + + if ( uSymAddr < uKernelStart + && (pThis->f64Bit ? LNX64_VALID_ADDRESS(uSymAddr) : LNX32_VALID_ADDRESS(uSymAddr)) + && uKernelStart - uSymAddr < LNX_MAX_KERNEL_SIZE) + uKernelStart = uSymAddr; + } + + for (i = pThis->cKernelSymbols - 1; i > 0; i--) + { + RTGCUINTPTR uSymAddr = dbgDiggerLinuxConvOffsetToAddr(pThis, pai32Offsets[i]); + + if ( uSymAddr > uKernelEnd + && (pThis->f64Bit ? LNX64_VALID_ADDRESS(uSymAddr) : LNX32_VALID_ADDRESS(uSymAddr)) + && uSymAddr - uKernelEnd < LNX_MAX_KERNEL_SIZE) + uKernelEnd = uSymAddr; + + /* Store the offset from the derived kernel start address. */ + pauSymOff[i] = uSymAddr - uKernelStart; + } + + RTGCUINTPTR cbKernel = uKernelEnd - uKernelStart; + pThis->cbKernel = (uint32_t)cbKernel; + pVMM->pfnDBGFR3AddrFromFlat(pUVM, &pThis->AddrKernelBase, uKernelStart); + Log(("dbgDiggerLinuxLoadKernelSymbolsRelative: uKernelStart=%RGv cbKernel=%#x\n", uKernelStart, cbKernel)); + + rc = dbgDiggerLinuxLoadKernelSymbolsWorker(pUVM, pVMM, pThis, uKernelStart, cbKernel, pauSymOff); + if (RT_FAILURE(rc)) + Log(("dbgDiggerLinuxLoadKernelSymbolsRelative: Loading symbols from given offset table failed: %Rrc\n", rc)); + RTMemTmpFree(pauSymOff); + } + else + Log(("dbgDiggerLinuxLoadKernelSymbolsRelative: Reading symbol addresses at %RGv failed: %Rrc\n", + pThis->AddrKernelAddresses.FlatPtr, rc)); + RTMemFree(pai32Offsets); + + return rc; +} + + +/** + * Loads the kernel symbols. + * + * @returns VBox status code. + * @param pUVM The user mode VM handle. + * @param pVMM The VMM function table. + * @param pThis The Linux digger data. + */ +static int dbgDiggerLinuxLoadKernelSymbols(PUVM pUVM, PCVMMR3VTABLE pVMM, PDBGDIGGERLINUX pThis) +{ + /* + * First the kernel itself. + */ + if (pThis->fRelKrnlAddr) + return dbgDiggerLinuxLoadKernelSymbolsRelative(pUVM, pVMM, pThis); + return dbgDiggerLinuxLoadKernelSymbolsAbsolute(pUVM, pVMM, pThis); +} + + +/* + * The module structure changed it was easier to produce different code for + * each version of the structure. The C preprocessor rules! + */ +#define LNX_TEMPLATE_HEADER "DBGPlugInLinuxModuleCodeTmpl.cpp.h" + +#define LNX_BIT_SUFFIX _amd64 +#define LNX_PTR_T uint64_t +#define LNX_64BIT 1 +#include "DBGPlugInLinuxModuleVerTmpl.cpp.h" + +#define LNX_BIT_SUFFIX _x86 +#define LNX_PTR_T uint32_t +#define LNX_64BIT 0 +#include "DBGPlugInLinuxModuleVerTmpl.cpp.h" + +#undef LNX_TEMPLATE_HEADER + +static const struct +{ + uint32_t uVersion; + bool f64Bit; + uint64_t (*pfnProcessModule)(PDBGDIGGERLINUX pThis, PUVM pUVM, PCVMMR3VTABLE pVMM, PDBGFADDRESS pAddrModule); +} g_aModVersions[] = +{ +#define LNX_TEMPLATE_HEADER "DBGPlugInLinuxModuleTableEntryTmpl.cpp.h" + +#define LNX_BIT_SUFFIX _amd64 +#define LNX_64BIT 1 +#include "DBGPlugInLinuxModuleVerTmpl.cpp.h" + +#define LNX_BIT_SUFFIX _x86 +#define LNX_64BIT 0 +#include "DBGPlugInLinuxModuleVerTmpl.cpp.h" + +#undef LNX_TEMPLATE_HEADER +}; + + +/** + * Tries to find and process the module list. + * + * @returns VBox status code. + * @param pThis The Linux digger data. + * @param pUVM The user mode VM handle. + * @param pVMM The VMM function table. + */ +static int dbgDiggerLinuxLoadModules(PDBGDIGGERLINUX pThis, PUVM pUVM, PCVMMR3VTABLE pVMM) +{ + /* + * Locate the list head. + */ + RTDBGAS hAs = pVMM->pfnDBGFR3AsResolveAndRetain(pUVM, DBGF_AS_KERNEL); + RTDBGSYMBOL SymInfo; + int rc = RTDbgAsSymbolByName(hAs, "vmlinux!modules", &SymInfo, NULL); + RTDbgAsRelease(hAs); + if (RT_FAILURE(rc)) + return VERR_NOT_FOUND; + + if (RT_FAILURE(rc)) + { + LogRel(("dbgDiggerLinuxLoadModules: Failed to locate the module list (%Rrc).\n", rc)); + return VERR_NOT_FOUND; + } + + /* + * Read the list anchor. + */ + union + { + uint32_t volatile u32Pair[2]; + uint64_t u64Pair[2]; + } uListAnchor; + DBGFADDRESS Addr; + rc = pVMM->pfnDBGFR3MemRead(pUVM, 0 /*idCpu*/, pVMM->pfnDBGFR3AddrFromFlat(pUVM, &Addr, SymInfo.Value), + &uListAnchor, pThis->f64Bit ? sizeof(uListAnchor.u64Pair) : sizeof(uListAnchor.u32Pair)); + if (RT_FAILURE(rc)) + { + LogRel(("dbgDiggerLinuxLoadModules: Error reading list anchor at %RX64: %Rrc\n", SymInfo.Value, rc)); + return VERR_NOT_FOUND; + } + if (!pThis->f64Bit) + { + uListAnchor.u64Pair[1] = uListAnchor.u32Pair[1]; + ASMCompilerBarrier(); + uListAnchor.u64Pair[0] = uListAnchor.u32Pair[0]; + } + + if (pThis->uKrnlVer == 0) + { + LogRel(("dbgDiggerLinuxLoadModules: No valid kernel version given: %#x\n", pThis->uKrnlVer)); + return VERR_NOT_FOUND; + } + + /* + * Find the g_aModVersion entry that fits the best. + * ASSUMES strict descending order by bitcount and version. + */ + Assert(g_aModVersions[0].f64Bit == true); + unsigned i = 0; + if (!pThis->f64Bit) + while (i < RT_ELEMENTS(g_aModVersions) && g_aModVersions[i].f64Bit) + i++; + while ( i < RT_ELEMENTS(g_aModVersions) + && g_aModVersions[i].f64Bit == pThis->f64Bit + && pThis->uKrnlVer < g_aModVersions[i].uVersion) + i++; + if (i >= RT_ELEMENTS(g_aModVersions)) + { + LogRel(("dbgDiggerLinuxLoadModules: Failed to find anything matching version: %u.%u.%u\n", + pThis->uKrnlVerMaj, pThis->uKrnlVerMin, pThis->uKrnlVerBld)); + return VERR_NOT_FOUND; + } + + /* + * Walk the list. + */ + uint64_t uModAddr = uListAnchor.u64Pair[0]; + for (size_t iModule = 0; iModule < 4096 && uModAddr != SymInfo.Value && uModAddr != 0; iModule++) + uModAddr = g_aModVersions[i].pfnProcessModule(pThis, pUVM, pVMM, pVMM->pfnDBGFR3AddrFromFlat(pUVM, &Addr, uModAddr)); + + return VINF_SUCCESS; +} + + +/** + * Checks if there is a likely kallsyms_names fragment at pHitAddr. + * + * @returns true if it's a likely fragment, false if not. + * @param pUVM The user mode VM handle. + * @param pVMM The VMM function table. + * @param pHitAddr The address where paNeedle was found. + * @param pabNeedle The fragment we've been searching for. + * @param cbNeedle The length of the fragment. + */ +static bool dbgDiggerLinuxIsLikelyNameFragment(PUVM pUVM, PCVMMR3VTABLE pVMM, PCDBGFADDRESS pHitAddr, + uint8_t const *pabNeedle, uint8_t cbNeedle) +{ + /* + * Examples of lead and tail bytes of our choosen needle in a randomly + * picked kernel: + * k o b j + * 22 6b 6f 62 6a aa + * fc 6b 6f 62 6a aa + * 82 6b 6f 62 6a 5f - ascii trail byte (_). + * ee 6b 6f 62 6a aa + * fc 6b 6f 62 6a 5f - ascii trail byte (_). + * 0a 74 6b 6f 62 6a 5f ea - ascii lead (t) and trail (_) bytes. + * 0b 54 6b 6f 62 6a aa - ascii lead byte (T). + * ... omitting 29 samples similar to the last two ... + * d8 6b 6f 62 6a aa + * d8 6b 6f 62 6a aa + * d8 6b 6f 62 6a aa + * d8 6b 6f 62 6a aa + * f9 5f 6b 6f 62 6a 5f 94 - ascii lead and trail bytes (_) + * f9 5f 6b 6f 62 6a 0c - ascii lead byte (_). + * fd 6b 6f 62 6a 0f + * ... enough. + */ + uint8_t abBuf[32]; + DBGFADDRESS ReadAddr = *pHitAddr; + pVMM->pfnDBGFR3AddrSub(&ReadAddr, 2); + int rc = pVMM->pfnDBGFR3MemRead(pUVM, 0 /*idCpu*/, &ReadAddr, abBuf, 2 + cbNeedle + 2); + if (RT_SUCCESS(rc)) + { + if (memcmp(&abBuf[2], pabNeedle, cbNeedle) == 0) /* paranoia */ + { + uint8_t const bLead = abBuf[1] == '_' || abBuf[1] == 'T' || abBuf[1] == 't' ? abBuf[0] : abBuf[1]; + uint8_t const offTail = 2 + cbNeedle; + uint8_t const bTail = abBuf[offTail] == '_' ? abBuf[offTail] : abBuf[offTail + 1]; + if ( bLead >= 1 && (bLead < 0x20 || bLead >= 0x80) + && bTail >= 1 && (bTail < 0x20 || bTail >= 0x80)) + return true; + Log(("dbgDiggerLinuxIsLikelyNameFragment: failed at %RGv: bLead=%#x bTail=%#x (offTail=%#x)\n", + pHitAddr->FlatPtr, bLead, bTail, offTail)); + } + else + Log(("dbgDiggerLinuxIsLikelyNameFragment: failed at %RGv: Needle changed!\n", pHitAddr->FlatPtr)); + } + else + Log(("dbgDiggerLinuxIsLikelyNameFragment: failed at %RGv: %Rrc\n", pHitAddr->FlatPtr, rc)); + + return false; +} + +/** + * Tries to find and load the kernel symbol table with the given needle. + * + * @returns VBox status code. + * @param pThis The Linux digger data. + * @param pUVM The user mode VM handle. + * @param pVMM The VMM function table. + * @param pabNeedle The needle to use for searching. + * @param cbNeedle Size of the needle in bytes. + */ +static int dbgDiggerLinuxFindSymbolTableFromNeedle(PDBGDIGGERLINUX pThis, PUVM pUVM, PCVMMR3VTABLE pVMM, + uint8_t const *pabNeedle, uint8_t cbNeedle) +{ + /* + * Go looking for the kallsyms table. If it's there, it will be somewhere + * after the linux_banner symbol, so use it for starting the search. + */ + int rc = VINF_SUCCESS; + DBGFADDRESS CurAddr = pThis->AddrLinuxBanner; + uint32_t cbLeft = LNX_MAX_KERNEL_SIZE; + while (cbLeft > 4096) + { + DBGFADDRESS HitAddr; + rc = pVMM->pfnDBGFR3MemScan(pUVM, 0 /*idCpu*/, &CurAddr, cbLeft, 1 /*uAlign*/, + pabNeedle, cbNeedle, &HitAddr); + if (RT_FAILURE(rc)) + break; + if (dbgDiggerLinuxIsLikelyNameFragment(pUVM, pVMM, &HitAddr, pabNeedle, cbNeedle)) + { + /* There will be another hit near by. */ + pVMM->pfnDBGFR3AddrAdd(&HitAddr, 1); + rc = pVMM->pfnDBGFR3MemScan(pUVM, 0 /*idCpu*/, &HitAddr, LNX_MAX_KALLSYMS_NAMES_SIZE, 1 /*uAlign*/, + pabNeedle, cbNeedle, &HitAddr); + if ( RT_SUCCESS(rc) + && dbgDiggerLinuxIsLikelyNameFragment(pUVM, pVMM, &HitAddr, pabNeedle, cbNeedle)) + { + /* + * We've got a very likely candidate for a location inside kallsyms_names. + * Try find the start of it, that is to say, try find kallsyms_num_syms. + * kallsyms_num_syms is aligned on sizeof(unsigned long) boundrary + */ + rc = dbgDiggerLinuxFindStartOfNamesAndSymbolCount(pUVM, pVMM, pThis, &HitAddr); + if (RT_SUCCESS(rc)) + rc = dbgDiggerLinuxFindEndOfNamesAndMore(pUVM, pVMM, pThis, &HitAddr); + if (RT_SUCCESS(rc)) + rc = dbgDiggerLinuxFindTokenIndex(pUVM, pVMM, pThis); + if (RT_SUCCESS(rc)) + rc = dbgDiggerLinuxLoadKernelSymbols(pUVM, pVMM, pThis); + if (RT_SUCCESS(rc)) + { + rc = dbgDiggerLinuxLoadModules(pThis, pUVM, pVMM); + break; + } + } + } + + /* + * Advance. + */ + RTGCUINTPTR cbDistance = HitAddr.FlatPtr - CurAddr.FlatPtr + cbNeedle; + if (RT_UNLIKELY(cbDistance >= cbLeft)) + { + Log(("dbgDiggerLinuxInit: Failed to find kallsyms\n")); + break; + } + cbLeft -= cbDistance; + pVMM->pfnDBGFR3AddrAdd(&CurAddr, cbDistance); + } + + return rc; +} + +/** + * Skips whitespace and comments in the given config returning the pointer + * to the first non whitespace character. + * + * @returns Pointer to the first non whitespace character or NULL if the end + * of the string was reached. + * @param pszCfg The config string. + */ +static const char *dbgDiggerLinuxCfgSkipWhitespace(const char *pszCfg) +{ + do + { + while ( *pszCfg != '\0' + && ( RT_C_IS_SPACE(*pszCfg) + || *pszCfg == '\n')) + pszCfg++; + + /* Do we have a comment? Skip it. */ + if (*pszCfg == '#') + { + while ( *pszCfg != '\n' + && *pszCfg != '\0') + pszCfg++; + } + } while ( *pszCfg != '\0' + && ( RT_C_IS_SPACE(*pszCfg) + || *pszCfg == '\n' + || *pszCfg == '#')); + + return pszCfg; +} + +/** + * Parses an identifier at the given position. + * + * @returns VBox status code. + * @param pszCfg The config data. + * @param ppszCfgNext Where to store the pointer to the data following the identifier. + * @param ppszIde Where to store the pointer to the identifier on success. + * Free with RTStrFree(). + */ +static int dbgDiggerLinuxCfgParseIde(const char *pszCfg, const char **ppszCfgNext, char **ppszIde) +{ + int rc = VINF_SUCCESS; + size_t cchIde = 0; + + while ( *pszCfg != '\0' + && ( RT_C_IS_ALNUM(*pszCfg) + || *pszCfg == '_')) + { + cchIde++; + pszCfg++; + } + + if (cchIde) + { + *ppszIde = RTStrDupN(pszCfg - cchIde, cchIde); + if (!*ppszIde) + rc = VERR_NO_STR_MEMORY; + } + + *ppszCfgNext = pszCfg; + return rc; +} + +/** + * Parses a value for a config item. + * + * @returns VBox status code. + * @param pszCfg The config data. + * @param ppszCfgNext Where to store the pointer to the data following the identifier. + * @param ppCfgItem Where to store the created config item on success. + */ +static int dbgDiggerLinuxCfgParseVal(const char *pszCfg, const char **ppszCfgNext, + PDBGDIGGERLINUXCFGITEM *ppCfgItem) +{ + int rc = VINF_SUCCESS; + PDBGDIGGERLINUXCFGITEM pCfgItem = NULL; + + if (RT_C_IS_DIGIT(*pszCfg) || *pszCfg == '-') + { + /* Parse the number. */ + int64_t i64Num; + rc = RTStrToInt64Ex(pszCfg, (char **)ppszCfgNext, 0, &i64Num); + if ( RT_SUCCESS(rc) + || rc == VWRN_TRAILING_CHARS + || rc == VWRN_TRAILING_SPACES) + { + pCfgItem = (PDBGDIGGERLINUXCFGITEM)RTMemAllocZ(sizeof(DBGDIGGERLINUXCFGITEM)); + if (pCfgItem) + { + pCfgItem->enmType = DBGDIGGERLINUXCFGITEMTYPE_NUMBER; + pCfgItem->u.i64Num = i64Num; + } + else + rc = VERR_NO_MEMORY; + } + } + else if (*pszCfg == '\"') + { + /* Parse a string. */ + const char *pszCfgCur = pszCfg + 1; + while ( *pszCfgCur != '\0' + && *pszCfgCur != '\"') + pszCfgCur++; + + if (*pszCfgCur == '\"') + { + pCfgItem = (PDBGDIGGERLINUXCFGITEM)RTMemAllocZ(RT_UOFFSETOF_DYN(DBGDIGGERLINUXCFGITEM, + u.aszString[pszCfgCur - pszCfg + 1])); + if (pCfgItem) + { + pCfgItem->enmType = DBGDIGGERLINUXCFGITEMTYPE_STRING; + RTStrCopyEx(&pCfgItem->u.aszString[0], pszCfgCur - pszCfg + 1, pszCfg, pszCfgCur - pszCfg); + *ppszCfgNext = pszCfgCur + 1; + } + else + rc = VERR_NO_MEMORY; + } + else + rc = VERR_INVALID_STATE; + } + else if ( *pszCfg == 'y' + || *pszCfg == 'm') + { + /* Included or module. */ + pCfgItem = (PDBGDIGGERLINUXCFGITEM)RTMemAllocZ(sizeof(DBGDIGGERLINUXCFGITEM)); + if (pCfgItem) + { + pCfgItem->enmType = DBGDIGGERLINUXCFGITEMTYPE_FLAG; + pCfgItem->u.fModule = *pszCfg == 'm'; + } + else + rc = VERR_NO_MEMORY; + pszCfg++; + *ppszCfgNext = pszCfg; + } + else + rc = VERR_INVALID_STATE; + + if (RT_SUCCESS(rc)) + *ppCfgItem = pCfgItem; + else if (pCfgItem) + RTMemFree(pCfgItem); + + return rc; +} + +/** + * Parses the given kernel config and creates the config database. + * + * @returns VBox status code + * @param pThis The Linux digger data. + * @param pszCfg The config string. + */ +static int dbgDiggerLinuxCfgParse(PDBGDIGGERLINUX pThis, const char *pszCfg) +{ + int rc = VINF_SUCCESS; + + /* + * The config is a text file with the following elements: + * # starts a comment which goes till the end of the line + * <Ide>=<val> where <Ide> is an identifier consisting of + * alphanumerical characters (including _) + * <val> denotes the value for the identifier and can have the following + * formats: + * (-)[0-9]* for numbers + * "..." for a string value + * m when a feature is enabled as a module + * y when a feature is enabled + * Newlines are used as a separator between values and mark the end + * of a comment + */ + const char *pszCfgCur = pszCfg; + while ( RT_SUCCESS(rc) + && *pszCfgCur != '\0') + { + /* Start skipping the whitespace. */ + pszCfgCur = dbgDiggerLinuxCfgSkipWhitespace(pszCfgCur); + if ( pszCfgCur + && *pszCfgCur != '\0') + { + char *pszIde = NULL; + /* Must be an identifier, parse it. */ + rc = dbgDiggerLinuxCfgParseIde(pszCfgCur, &pszCfgCur, &pszIde); + if (RT_SUCCESS(rc)) + { + /* + * Skip whitespace again (shouldn't be required because = follows immediately + * in the observed configs). + */ + pszCfgCur = dbgDiggerLinuxCfgSkipWhitespace(pszCfgCur); + if ( pszCfgCur + && *pszCfgCur == '=') + { + pszCfgCur++; + pszCfgCur = dbgDiggerLinuxCfgSkipWhitespace(pszCfgCur); + if ( pszCfgCur + && *pszCfgCur != '\0') + { + /* Get the value. */ + PDBGDIGGERLINUXCFGITEM pCfgItem = NULL; + rc = dbgDiggerLinuxCfgParseVal(pszCfgCur, &pszCfgCur, &pCfgItem); + if (RT_SUCCESS(rc)) + { + pCfgItem->Core.pszString = pszIde; + bool fRc = RTStrSpaceInsert(&pThis->hCfgDb, &pCfgItem->Core); + if (!fRc) + { + RTStrFree(pszIde); + RTMemFree(pCfgItem); + rc = VERR_INVALID_STATE; + } + } + } + else + rc = VERR_EOF; + } + else + rc = VERR_INVALID_STATE; + } + + if (RT_FAILURE(rc)) + RTStrFree(pszIde); + } + else + break; /* Reached the end of the config. */ + } + + if (RT_FAILURE(rc)) + dbgDiggerLinuxCfgDbDestroy(pThis); + + return rc; +} + +/** + * Decompresses the given config and validates the UTF-8 encoding. + * + * @returns VBox status code. + * @param pbCfgComp The compressed config. + * @param cbCfgComp Size of the compressed config. + * @param ppszCfg Where to store the pointer to the decompressed config + * on success. + */ +static int dbgDiggerLinuxCfgDecompress(const uint8_t *pbCfgComp, size_t cbCfgComp, char **ppszCfg) +{ + int rc = VINF_SUCCESS; + RTVFSIOSTREAM hVfsIos = NIL_RTVFSIOSTREAM; + + rc = RTVfsIoStrmFromBuffer(RTFILE_O_READ, pbCfgComp, cbCfgComp, &hVfsIos); + if (RT_SUCCESS(rc)) + { + RTVFSIOSTREAM hVfsIosDecomp = NIL_RTVFSIOSTREAM; + rc = RTZipGzipDecompressIoStream(hVfsIos, RTZIPGZIPDECOMP_F_ALLOW_ZLIB_HDR, &hVfsIosDecomp); + if (RT_SUCCESS(rc)) + { + char *pszCfg = NULL; + size_t cchCfg = 0; + size_t cbRead = 0; + + do + { + uint8_t abBuf[_64K]; + rc = RTVfsIoStrmRead(hVfsIosDecomp, abBuf, sizeof(abBuf), true /*fBlocking*/, &cbRead); + if (rc == VINF_EOF && cbRead == 0) + rc = VINF_SUCCESS; + if ( RT_SUCCESS(rc) + && cbRead > 0) + { + /* Append data. */ + char *pszCfgNew = pszCfg; + rc = RTStrRealloc(&pszCfgNew, cchCfg + cbRead + 1); + if (RT_SUCCESS(rc)) + { + pszCfg = pszCfgNew; + memcpy(pszCfg + cchCfg, &abBuf[0], cbRead); + cchCfg += cbRead; + pszCfg[cchCfg] = '\0'; /* Enforce string termination. */ + } + } + } while (RT_SUCCESS(rc) && cbRead > 0); + + if (RT_SUCCESS(rc)) + *ppszCfg = pszCfg; + else if (RT_FAILURE(rc) && pszCfg) + RTStrFree(pszCfg); + + RTVfsIoStrmRelease(hVfsIosDecomp); + } + RTVfsIoStrmRelease(hVfsIos); + } + + return rc; +} + +/** + * Reads and decodes the compressed kernel config. + * + * @returns VBox status code. + * @param pThis The Linux digger data. + * @param pUVM The user mode VM handle. + * @param pVMM The VMM function table. + * @param pAddrStart The start address of the compressed config. + * @param cbCfgComp The size of the compressed config. + */ +static int dbgDiggerLinuxCfgDecode(PDBGDIGGERLINUX pThis, PUVM pUVM, PCVMMR3VTABLE pVMM, + PCDBGFADDRESS pAddrStart, size_t cbCfgComp) +{ + int rc = VINF_SUCCESS; + uint8_t *pbCfgComp = (uint8_t *)RTMemTmpAlloc(cbCfgComp); + if (!pbCfgComp) + return VERR_NO_MEMORY; + + rc = pVMM->pfnDBGFR3MemRead(pUVM, 0 /*idCpu*/, pAddrStart, pbCfgComp, cbCfgComp); + if (RT_SUCCESS(rc)) + { + char *pszCfg = NULL; + rc = dbgDiggerLinuxCfgDecompress(pbCfgComp, cbCfgComp, &pszCfg); + if (RT_SUCCESS(rc)) + { + if (RTStrIsValidEncoding(pszCfg)) + rc = dbgDiggerLinuxCfgParse(pThis, pszCfg); + else + rc = VERR_INVALID_UTF8_ENCODING; + RTStrFree(pszCfg); + } + } + + RTMemFree(pbCfgComp); + return rc; +} + +/** + * Tries to find the compressed kernel config in the kernel address space + * and sets up the config database. + * + * @returns VBox status code. + * @param pThis The Linux digger data. + * @param pUVM The user mode VM handle. + * @param pVMM The VMM function table. + */ +static int dbgDiggerLinuxCfgFind(PDBGDIGGERLINUX pThis, PUVM pUVM, PCVMMR3VTABLE pVMM) +{ + /* + * Go looking for the IKCFG_ST string which indicates the start + * of the compressed config file. + */ + static const uint8_t s_abCfgNeedleStart[] = "IKCFG_ST"; + static const uint8_t s_abCfgNeedleEnd[] = "IKCFG_ED"; + int rc = VINF_SUCCESS; + DBGFADDRESS CurAddr = pThis->AddrLinuxBanner; + uint32_t cbLeft = LNX_MAX_KERNEL_SIZE; + while (cbLeft > 4096) + { + DBGFADDRESS HitAddrStart; + rc = pVMM->pfnDBGFR3MemScan(pUVM, 0 /*idCpu*/, &CurAddr, cbLeft, 1 /*uAlign*/, + s_abCfgNeedleStart, sizeof(s_abCfgNeedleStart) - 1, &HitAddrStart); + if (RT_FAILURE(rc)) + break; + + /* Check for the end marker which shouldn't be that far away. */ + pVMM->pfnDBGFR3AddrAdd(&HitAddrStart, sizeof(s_abCfgNeedleStart) - 1); + DBGFADDRESS HitAddrEnd; + rc = pVMM->pfnDBGFR3MemScan(pUVM, 0 /* idCpu */, &HitAddrStart, LNX_MAX_COMPRESSED_CFG_SIZE, + 1 /* uAlign */, s_abCfgNeedleEnd, sizeof(s_abCfgNeedleEnd) - 1, &HitAddrEnd); + if (RT_SUCCESS(rc)) + { + /* Allocate a buffer to hold the compressed data between the markers and fetch it. */ + RTGCUINTPTR cbCfg = HitAddrEnd.FlatPtr - HitAddrStart.FlatPtr; + Assert(cbCfg == (size_t)cbCfg); + rc = dbgDiggerLinuxCfgDecode(pThis, pUVM, pVMM, &HitAddrStart, cbCfg); + if (RT_SUCCESS(rc)) + break; + } + + /* + * Advance. + */ + RTGCUINTPTR cbDistance = HitAddrStart.FlatPtr - CurAddr.FlatPtr + sizeof(s_abCfgNeedleStart) - 1; + if (RT_UNLIKELY(cbDistance >= cbLeft)) + { + LogFunc(("Failed to find compressed kernel config\n")); + break; + } + cbLeft -= cbDistance; + pVMM->pfnDBGFR3AddrAdd(&CurAddr, cbDistance); + } + + return rc; +} + +/** + * Probes for a Linux kernel starting at the given address. + * + * @returns Flag whether something which looks like a valid Linux kernel was found. + * @param pThis The Linux digger data. + * @param pUVM The user mode VM handle. + * @param pVMM The VMM function table. + * @param uAddrStart The address to start scanning at. + * @param cbScan How much to scan. + */ +static bool dbgDiggerLinuxProbeWithAddr(PDBGDIGGERLINUX pThis, PUVM pUVM, PCVMMR3VTABLE pVMM, + RTGCUINTPTR uAddrStart, size_t cbScan) +{ + /* + * Look for "Linux version " at the start of the rodata segment. + * Hope that this comes before any message buffer or other similar string. + */ + DBGFADDRESS KernelAddr; + pVMM->pfnDBGFR3AddrFromFlat(pUVM, &KernelAddr, uAddrStart); + DBGFADDRESS HitAddr; + int rc = pVMM->pfnDBGFR3MemScan(pUVM, 0, &KernelAddr, cbScan, 1, + g_abLinuxVersion, sizeof(g_abLinuxVersion) - 1, &HitAddr); + if (RT_SUCCESS(rc)) + { + char szTmp[128]; + char const *pszX = &szTmp[sizeof(g_abLinuxVersion) - 1]; + rc = pVMM->pfnDBGFR3MemReadString(pUVM, 0, &HitAddr, szTmp, sizeof(szTmp)); + if ( RT_SUCCESS(rc) + && ( ( pszX[0] == '2' /* 2.x.y with x in {0..6} */ + && pszX[1] == '.' + && pszX[2] >= '0' + && pszX[2] <= '6') + || ( pszX[0] >= '3' /* 3.x, 4.x, ... 9.x */ + && pszX[0] <= '9' + && pszX[1] == '.' + && pszX[2] >= '0' + && pszX[2] <= '9') + ) + ) + { + pThis->AddrKernelBase = KernelAddr; + pThis->AddrLinuxBanner = HitAddr; + return true; + } + } + + return false; +} + +/** + * Probes for a Linux kernel which has KASLR enabled. + * + * @returns Flag whether a possible candidate location was found. + * @param pThis The Linux digger data. + * @param pUVM The user mode VM handle. + * @param pVMM The VMM function table. + */ +static bool dbgDiggerLinuxProbeKaslr(PDBGDIGGERLINUX pThis, PUVM pUVM, PCVMMR3VTABLE pVMM) +{ + /** + * With KASLR the kernel is loaded at a different address at each boot making detection + * more difficult for us. + * + * The randomization is done in arch/x86/boot/compressed/kaslr.c:choose_random_location() (as of Nov 2017). + * At the end of the method a random offset is chosen using find_random_virt_addr() which is added to the + * kernel map start in the caller (the start of the kernel depends on the bit size, see LNX32_KERNEL_ADDRESS_START + * and LNX64_KERNEL_ADDRESS_START for 32bit and 64bit kernels respectively). + * The lowest offset possible is LOAD_PHYSICAL_ADDR which is defined in arch/x86/include/asm/boot.h + * using CONFIG_PHYSICAL_START aligned to CONFIG_PHYSICAL_ALIGN. + * The default CONFIG_PHYSICAL_START and CONFIG_PHYSICAL_ALIGN are both 0x1000000 no matter whether a 32bit + * or a 64bit kernel is used. So the lowest offset to the kernel start address is 0x1000000. + * The find_random_virt_addr() the number of possible slots where the kernel can be placed based on the image size + * is calculated using the following formula: + * cSlots = ((KERNEL_IMAGE_SIZE - 0x1000000 (minimum) - image_size) / 0x1000000 (CONFIG_PHYSICAL_ALIGN)) + 1 + * + * KERNEL_IMAGE_SIZE is 1GB for 64bit kernels and 512MB for 32bit kernels, so the maximum number of slots (resulting + * in the largest possible offset) can be achieved when image_size (which contains the real size of the kernel image + * which is unknown for us) goes to 0 and a 1GB KERNEL_IMAGE_SIZE is assumed. With that the biggest cSlots which can be + * achieved is 64. The chosen random offset is taken from a random long integer using kaslr_get_random_long() modulo the + * number of slots which selects a slot between 0 and 63. The final offset is calculated using: + * offAddr = random_addr * 0x1000000 (CONFIG_PHYSICAL_ALIGN) + 0x1000000 (minimum) + * + * So the highest offset the kernel can start is 0x40000000 which is 1GB (plus the maximum kernel size we defined). + */ + if (dbgDiggerLinuxProbeWithAddr(pThis, pUVM, pVMM, LNX64_KERNEL_ADDRESS_START, _1G + LNX_MAX_KERNEL_SIZE)) + return true; + + /* + * 32bit variant, makes sure we don't exceed the 4GB address space or DBGFR3MemScan() returns VERR_DBGF_MEM_NOT_FOUND immediately + * without searching the remainder of the address space. + * + * The default split is 3GB userspace and 1GB kernel, so we just search the entire upper 1GB kernel space. + */ + if (dbgDiggerLinuxProbeWithAddr(pThis, pUVM, pVMM, LNX32_KERNEL_ADDRESS_START, _4G - LNX32_KERNEL_ADDRESS_START)) + return true; + + return false; +} + +/** + * @copydoc DBGFOSREG::pfnInit + */ +static DECLCALLBACK(int) dbgDiggerLinuxInit(PUVM pUVM, PCVMMR3VTABLE pVMM, void *pvData) +{ + PDBGDIGGERLINUX pThis = (PDBGDIGGERLINUX)pvData; + Assert(!pThis->fValid); + + char szVersion[256] = "Linux version 4.19.0"; + int rc = pVMM->pfnDBGFR3MemReadString(pUVM, 0, &pThis->AddrLinuxBanner, &szVersion[0], sizeof(szVersion)); + if (RT_SUCCESS(rc)) + { + /* + * Get a numerical version number. + */ + const char *pszVersion = szVersion; + while (*pszVersion && !RT_C_IS_DIGIT(*pszVersion)) + pszVersion++; + + size_t offVersion = 0; + uint32_t uMajor = 0; + while (pszVersion[offVersion] && RT_C_IS_DIGIT(pszVersion[offVersion])) + uMajor = uMajor * 10 + pszVersion[offVersion++] - '0'; + + if (pszVersion[offVersion] == '.') + offVersion++; + + uint32_t uMinor = 0; + while (pszVersion[offVersion] && RT_C_IS_DIGIT(pszVersion[offVersion])) + uMinor = uMinor * 10 + pszVersion[offVersion++] - '0'; + + if (pszVersion[offVersion] == '.') + offVersion++; + + uint32_t uBuild = 0; + while (pszVersion[offVersion] && RT_C_IS_DIGIT(pszVersion[offVersion])) + uBuild = uBuild * 10 + pszVersion[offVersion++] - '0'; + + pThis->uKrnlVer = LNX_MK_VER(uMajor, uMinor, uBuild); + pThis->uKrnlVerMaj = uMajor; + pThis->uKrnlVerMin = uMinor; + pThis->uKrnlVerBld = uBuild; + if (pThis->uKrnlVer == 0) + LogRel(("dbgDiggerLinuxInit: Failed to parse version string: %s\n", pszVersion)); + } + + /* + * Assume 64-bit kernels all live way beyond 32-bit address space. + */ + pThis->f64Bit = pThis->AddrLinuxBanner.FlatPtr > UINT32_MAX; + pThis->fRelKrnlAddr = false; + + pThis->hCfgDb = NULL; + + /* + * Try to find the compressed kernel config and parse it before we try + * to get the symbol table, the config database is required to select + * the method to use. + */ + rc = dbgDiggerLinuxCfgFind(pThis, pUVM, pVMM); + if (RT_FAILURE(rc)) + LogFlowFunc(("Failed to find kernel config (%Rrc), no config database available\n", rc)); + + static const uint8_t s_abNeedle[] = "kobj"; + rc = dbgDiggerLinuxFindSymbolTableFromNeedle(pThis, pUVM, pVMM, s_abNeedle, sizeof(s_abNeedle) - 1); + if (RT_FAILURE(rc)) + { + /* Try alternate needle (seen on older x86 Linux kernels). */ + static const uint8_t s_abNeedleAlt[] = "kobjec"; + rc = dbgDiggerLinuxFindSymbolTableFromNeedle(pThis, pUVM, pVMM, s_abNeedleAlt, sizeof(s_abNeedleAlt) - 1); + if (RT_FAILURE(rc)) + { + static const uint8_t s_abNeedleOSuseX86[] = "nmi"; /* OpenSuSe 10.2 x86 */ + rc = dbgDiggerLinuxFindSymbolTableFromNeedle(pThis, pUVM, pVMM, s_abNeedleOSuseX86, sizeof(s_abNeedleOSuseX86) - 1); + } + } + + pThis->fValid = true; + return VINF_SUCCESS; +} + + +/** + * @copydoc DBGFOSREG::pfnProbe + */ +static DECLCALLBACK(bool) dbgDiggerLinuxProbe(PUVM pUVM, PCVMMR3VTABLE pVMM, void *pvData) +{ + PDBGDIGGERLINUX pThis = (PDBGDIGGERLINUX)pvData; + + for (unsigned i = 0; i < RT_ELEMENTS(g_au64LnxKernelAddresses); i++) + { + if (dbgDiggerLinuxProbeWithAddr(pThis, pUVM, pVMM, g_au64LnxKernelAddresses[i], LNX_MAX_KERNEL_SIZE)) + return true; + } + + /* Maybe the kernel uses KASLR. */ + if (dbgDiggerLinuxProbeKaslr(pThis, pUVM, pVMM)) + return true; + + return false; +} + + +/** + * @copydoc DBGFOSREG::pfnDestruct + */ +static DECLCALLBACK(void) dbgDiggerLinuxDestruct(PUVM pUVM, PCVMMR3VTABLE pVMM, void *pvData) +{ + RT_NOREF(pUVM, pVMM, pvData); +} + + +/** + * @copydoc DBGFOSREG::pfnConstruct + */ +static DECLCALLBACK(int) dbgDiggerLinuxConstruct(PUVM pUVM, PCVMMR3VTABLE pVMM, void *pvData) +{ + RT_NOREF(pUVM, pVMM); + PDBGDIGGERLINUX pThis = (PDBGDIGGERLINUX)pvData; + pThis->IDmesg.u32Magic = DBGFOSIDMESG_MAGIC; + pThis->IDmesg.pfnQueryKernelLog = dbgDiggerLinuxIDmsg_QueryKernelLog; + pThis->IDmesg.u32EndMagic = DBGFOSIDMESG_MAGIC; + + return VINF_SUCCESS; +} + + +const DBGFOSREG g_DBGDiggerLinux = +{ + /* .u32Magic = */ DBGFOSREG_MAGIC, + /* .fFlags = */ 0, + /* .cbData = */ sizeof(DBGDIGGERLINUX), + /* .szName = */ "Linux", + /* .pfnConstruct = */ dbgDiggerLinuxConstruct, + /* .pfnDestruct = */ dbgDiggerLinuxDestruct, + /* .pfnProbe = */ dbgDiggerLinuxProbe, + /* .pfnInit = */ dbgDiggerLinuxInit, + /* .pfnRefresh = */ dbgDiggerLinuxRefresh, + /* .pfnTerm = */ dbgDiggerLinuxTerm, + /* .pfnQueryVersion = */ dbgDiggerLinuxQueryVersion, + /* .pfnQueryInterface = */ dbgDiggerLinuxQueryInterface, + /* .pfnStackUnwindAssist = */ dbgDiggerLinuxStackUnwindAssist, + /* .u32EndMagic = */ DBGFOSREG_MAGIC +}; + |