search for: ud2a

...ernel/paravirt.c b/arch/x86/kernel/paravirt.c index c2130aef3f9d..4f32a10979db 100644 --- a/arch/x86/kernel/paravirt.c +++ b/arch/x86/kernel/paravirt.c @@ -74,16 +74,6 @@ void __init default_banner(void) /* Undefined instruction for dealing with missing ops pointers. */ static const unsigned char ud2a[] = { 0x0f, 0x0b }; -unsigned paravirt_patch_nop(void) -{ - return 0; -} - -unsigned paravirt_patch_ignore(unsigned len) -{ - return len; -} - struct branch { unsigned char opcode; u32 delta; @@ -152,8 +142,7 @@ unsigned paravirt_patch_default(u8 type, u16 clobbers, void *insnbuf, /* If t...

...ernel/paravirt.c b/arch/x86/kernel/paravirt.c index c2130aef3f9d..4f32a10979db 100644 --- a/arch/x86/kernel/paravirt.c +++ b/arch/x86/kernel/paravirt.c @@ -74,16 +74,6 @@ void __init default_banner(void) /* Undefined instruction for dealing with missing ops pointers. */ static const unsigned char ud2a[] = { 0x0f, 0x0b }; -unsigned paravirt_patch_nop(void) -{ - return 0; -} - -unsigned paravirt_patch_ignore(unsigned len) -{ - return len; -} - struct branch { unsigned char opcode; u32 delta; @@ -152,8 +142,7 @@ unsigned paravirt_patch_default(u8 type, u16 clobbers, void *insnbuf, /* If t...

...ld do something like: if (type >= sizeof(pv_ops) / sizeof(void *)) return -EINVAL; opfunc = *((void **)&pv_ops + type); 124 unsigned ret; 125 126 if (opfunc == NULL) 127 /* If there's no function, patch it with a ud2a (BUG) */ 128 ret = paravirt_patch_insns(insn_buff, len, ud2a, ud2a+sizeof(ud2a)); 129 else if (opfunc == _paravirt_nop) 130 ret = 0; 131 regards, dan carpenter

> What about creating an msan interface DSO for the purposes of satisfying -z > defs? The executable will always be searched first, so the interface DSO > could be empty or full of ud2a. I like the idea. The ideal solution would probably be for the linkers to have a --not-needed command line option that prevents the .so from going in the DT_NEEDED, but one with just stubs is probably fine. Cheers, Rafael

Hi folks, With this series, the bulk of the work of pvops64 is done. Here, I integrate most of the paravirt.c and paravirt.h files, making them applicable to both architectures. CONFIG_PARAVIRT is _not_ present yet. Basically, this code is missing page table integration (patches currently being worked on by Jeremy). Enjoy

Hi folks, With this series, the bulk of the work of pvops64 is done. Here, I integrate most of the paravirt.c and paravirt.h files, making them applicable to both architectures. CONFIG_PARAVIRT is _not_ present yet. Basically, this code is missing page table integration (patches currently being worked on by Jeremy). Enjoy

...the kernels I use have the kdb patch installed. 1 kmem_cache_alloc (offset 0x125) get_unused_buffer_head journal_write_metadata_buffer journal_commit_transaction kjournald kernel_thread kmem_cache_alloc dis xchg %eax, (%ebx) cmp $0x5a2cf071, %eax (where did a value like that come from?) je .... ud2a - (looks like assembly for BUG()) One more thing I have seen is a problem with the following code do { new_bh = get_unused_buffer_head(0); if (!new_bh) { printk (KERN_NOTICE __FUNCTION__ ": ENOMEM...

...and can therefore be freely * converted back into a structure offset. + * + * For modules, generate a bogus BUG, which leaves enough space for + * the patch to a direct call to be made and eliminates the need to + * export paravirt_ops to modules. */ +#ifdef MODULE +#define PARAVIRT_CALL "ud2a; nop; nop; nop;" +#else #define PARAVIRT_CALL "call *(paravirt_ops+%c[paravirt_typenum]*4);" +#endif /* * These macros are intended to wrap calls into a paravirt_ops

...and can therefore be freely * converted back into a structure offset. + * + * For modules, generate a bogus BUG, which leaves enough space for + * the patch to a direct call to be made and eliminates the need to + * export paravirt_ops to modules. */ +#ifdef MODULE +#define PARAVIRT_CALL "ud2a; nop; nop; nop;" +#else #define PARAVIRT_CALL "call *(paravirt_ops+%c[paravirt_typenum]*4);" +#endif /* * These macros are intended to wrap calls into a paravirt_ops

...mit_timeout+0/10> Trace; c0108c2f <kernel_thread+1f/38> Trace; c0108c38 <kernel_thread+28/38> Code; c014ca15 <journal_write_metadata_buffer+4d/1d8> 00000000 <_EIP>: Code; c014ca15 <journal_write_metadata_buffer+4d/1d8> <===== 0: 0f 0b ud2a <===== Code; c014ca17 <journal_write_metadata_buffer+4f/1d8> 2: 83 c4 10 add $0x10,%esp Code; c014ca1a <journal_write_metadata_buffer+52/1d8> 5: 89 f6 mov %esi,%esi Code; c014ca1c <journal_write_metadata_buffer+54/1d8&gt...

I recently tried to enabled building llvm libraries with -Wl,-z,defs. The intention was to detect missing dependencies on ELF, so that we don't get into a situation where a .so builds, but the equivalent .dylib or .dll fails. This failed when building with msan because of undefined references to functions like __msan_memcpy. Unfortunately, I don't think elf linkers have a way of

...ore code, it also creates prologue/epilogue in otherwise leaf functions. Such mode may still be useful if for whatever reason we can not use SIGILL. Default (use ud2): 402ed5: 48 89 d8 mov %rbx,%rax << move the address to rax 402ed8: 0f 0b ud2a << crash 402eda: 52 push %rdx << encode is_write and size in the opcode (note: with a good disassembler and some work we can leave just ud2 or equivalent) -mllvm -asan-use-call 402ed5: 48 89 df mov %rbx,%rdi &...

...e_read_actor+0/e0> Trace; c01366e6 <filp_open+36/60> Trace; c0136a34 <sys_open+34/c0> Trace; c0106ffb <system_call+33/38> Code; c0165908 <journal_start+58/d0> 00000000 <_EIP>: Code; c0165908 <journal_start+58/d0> <===== 0: 0f 0b ud2a <===== Code; c016590a <journal_start+5a/d0> 2: 83 c4 14 add $0x14,%esp Code; c016590d <journal_start+5d/d0> 5: ff 43 08 incl 0x8(%ebx) Code; c0165910 <journal_start+60/d0> 8: eb 65 jmp 6f <_...

...ATIVE(irq_enable_sysexit, "sti; sysexit"); -DEF_NATIVE(read_cr2, "mov %cr2, %eax"); -DEF_NATIVE(write_cr3, "mov %eax, %cr3"); -DEF_NATIVE(read_cr3, "mov %cr3, %eax"); -DEF_NATIVE(clts, "clts"); -DEF_NATIVE(read_tsc, "rdtsc"); - -DEF_NATIVE(ud2a, "ud2a"); +#define DEF_NATIVE(ops, name, code) \ + extern const char start_##ops##_##name[], end_##ops##_##name[]; \ + asm("start_" #ops "_" #name ": " code "; end_" #ops "_" #name ":") + +DEF_NATIVE(pv_irq_ops, irq_disable,...

...ATIVE(irq_enable_sysexit, "sti; sysexit"); -DEF_NATIVE(read_cr2, "mov %cr2, %eax"); -DEF_NATIVE(write_cr3, "mov %eax, %cr3"); -DEF_NATIVE(read_cr3, "mov %cr3, %eax"); -DEF_NATIVE(clts, "clts"); -DEF_NATIVE(read_tsc, "rdtsc"); - -DEF_NATIVE(ud2a, "ud2a"); +#define DEF_NATIVE(ops, name, code) \ + extern const char start_##ops##_##name[], end_##ops##_##name[]; \ + asm("start_" #ops "_" #name ": " code "; end_" #ops "_" #name ":") + +DEF_NATIVE(pv_irq_ops, irq_disable,...

...ogue in otherwise leaf > functions. > Such mode may still be useful if for whatever reason we can not use > SIGILL. > > Default (use ud2): > 402ed5: 48 89 d8 mov %rbx,%rax << move the > address to rax > 402ed8: 0f 0b ud2a << crash > 402eda: 52 push %rdx << encode > is_write and size in the opcode > (note: with a good disassembler and some work we can leave just ud2 or > equivalent) > > -mllvm -asan-use-call > 402ed5: 48 89 df...

...mmit_timeout+0/10> Trace; c01072ee <arch_kernel_thread+2e/40> Trace; c0176a10 <kjournald+0/1e0> Code; c0174064 <journal_commit_transaction+924/11c0> 00000000 <_EIP>: Code; c0174064 <journal_commit_transaction+924/11c0> <===== 0: 0f 0b ud2a <===== Code; c0174066 <journal_commit_transaction+926/11c0> 2: f7 02 aa ff 29 c0 testl $0xc029ffaa,(%edx) Code; c017406c <journal_commit_transaction+92c/11c0> 8: e9 1b fe ff ff jmp fffffe28 <_EIP+0xfffffe28> Code; c0174071 <journal_co...

...kjournald+140/1d8> Trace; c01700d0 <commit_timeout+0/c> Trace; c010560f <kernel_thread+23/30> Code; c0170396 <journal_write_metadata_buffer+6a/2cc> 00000000 <_EIP>: Code; c0170396 <journal_write_metadata_buffer+6a/2cc> <===== 0: 0f 0b ud2a <===== Code; c0170398 <journal_write_metadata_buffer+6c/2cc> 2: 83 c4 08 add $0x8,%esp Code; c017039b <journal_write_metadata_buffer+6f/2cc> 5: 90 nop Code; c017039c <journal_write_metadata_buffer+70/2cc> 6:...

...e9 <END_OF_CODE+100088b1/????> Trace; ffff0018 <END_OF_CODE+2f780de0/????> Trace; c01056dc <kernel_thread+28/38> Code; c015822a <ext3_prepare_write+5e/1f4> 00000000 <_EIP>: Code; c015822a <ext3_prepare_write+5e/1f4> <===== 0: 0f 0b ud2a <===== Code; c015822c <ext3_prepare_write+60/1f4> 2: 83 c4 0c add $0xc,%esp Code; c015822f <ext3_prepare_write+63/1f4> 5: 90 nop Code; c0158230 <ext3_prepare_write+64/1f4> 6: 56 push...

...t 8:46 PM, Rafael Espíndola > <rafael.espindola at gmail.com> wrote: > >> What about creating an msan interface DSO for the purposes of > satisfying -z > >> defs? The executable will always be searched first, so the interface DSO > >> could be empty or full of ud2a. > > > > I like the idea. The ideal solution would probably be for the linkers > > to have a --not-needed command line option that prevents the .so from > > going in the DT_NEEDED, but one with just stubs is probably fine. > > But then the program would fail to start i...