kernel/linux.git/lib/string.c, branch v3.18.136

lib: make memzero_explicit more robust against dead store elimination

2018-11-10T15:39:19+00:00

[ Upstream commit 7829fb09a2b4268b30dd9bc782fa5ebee278b137 ] In commit 0b053c951829 ("lib: memzero_explicit: use barrier instead of OPTIMIZER_HIDE_VAR"), we made memzero_explicit() more robust in case LTO would decide to inline memzero_explicit() and eventually find out it could be elimiated as dead store. While using barrier() works well for the case of gcc, recent efforts from LLVMLinux people suggest to use llvm as an alternative to gcc, and there, Stephan found in a simple stand-alone user space example that llvm could nevertheless optimize and thus elimitate the memset(). A similar issue has been observed in the referenced llvm bug report, which is regarded as not-a-bug. Based on some experiments, icc is a bit special on its own, while it doesn't seem to eliminate the memset(), it could do so with an own implementation, and then result in similar findings as with llvm. The fix in this patch now works for all three compilers (also tested with more aggressive optimization levels). Arguably, in the current kernel tree it's more of a theoretical issue, but imho, it's better to be pedantic about it. It's clearly visible with gcc/llvm though, with the below code: if we would have used barrier() only here, llvm would have omitted clearing, not so with barrier_data() variant: static inline void memzero_explicit(void *s, size_t count) { memset(s, 0, count); barrier_data(s); } int main(void) { char buff[20]; memzero_explicit(buff, sizeof(buff)); return 0; } $ gcc -O2 test.c $ gdb a.out (gdb) disassemble main Dump of assembler code for function main: 0x0000000000400400 <+0>: lea -0x28(%rsp),%rax 0x0000000000400405 <+5>: movq $0x0,-0x28(%rsp) 0x000000000040040e <+14>: movq $0x0,-0x20(%rsp) 0x0000000000400417 <+23>: movl $0x0,-0x18(%rsp) 0x000000000040041f <+31>: xor %eax,%eax 0x0000000000400421 <+33>: retq End of assembler dump. $ clang -O2 test.c $ gdb a.out (gdb) disassemble main Dump of assembler code for function main: 0x00000000004004f0 <+0>: xorps %xmm0,%xmm0 0x00000000004004f3 <+3>: movaps %xmm0,-0x18(%rsp) 0x00000000004004f8 <+8>: movl $0x0,-0x8(%rsp) 0x0000000000400500 <+16>: lea -0x18(%rsp),%rax 0x0000000000400505 <+21>: xor %eax,%eax 0x0000000000400507 <+23>: retq End of assembler dump. As gcc, clang, but also icc defines __GNUC__, it's sufficient to define this in compiler-gcc.h only to be picked up. For a fallback or otherwise unsupported compiler, we define it as a barrier. Similarly, for ecc which does not support gcc inline asm. Reference: https://llvm.org/bugs/show_bug.cgi?id=15495 Reported-by: Stephan Mueller Tested-by: Stephan Mueller Signed-off-by: Daniel Borkmann Cc: Theodore Ts'o Cc: Stephan Mueller Cc: Hannes Frederic Sowa Cc: mancha security Cc: Mark Charlebois Cc: Behan Webster Signed-off-by: Herbert Xu Signed-off-by: Sasha Levin

strscpy: zero any trailing garbage bytes in the destination

2017-08-11T16:30:09+00:00

commit 990486c8af044f89bddfbde1d1cf9fde449bedbf upstream. It's possible that the destination can be shadowed in userspace (as, for example, the perf buffers are now). So we should take care not to leak data that could be inspected by userspace. Signed-off-by: Chris Metcalf Signed-off-by: Greg Kroah-Hartman

string: provide strscpy()

2017-08-11T16:30:09+00:00

commit 30035e45753b708e7d47a98398500ca005e02b86 upstream. The strscpy() API is intended to be used instead of strlcpy(), and instead of most uses of strncpy(). - Unlike strlcpy(), it doesn't read from memory beyond (src + size). - Unlike strlcpy() or strncpy(), the API provides an easy way to check for destination buffer overflow: an -E2BIG error return value. - The provided implementation is robust in the face of the source buffer being asynchronously changed during the copy, unlike the current implementation of strlcpy(). - Unlike strncpy(), the destination buffer will be NUL-terminated if the string in the source buffer is too long. - Also unlike strncpy(), the destination buffer will not be updated beyond the NUL termination, avoiding strncpy's behavior of zeroing the entire tail end of the destination buffer. (A memset() after the strscpy() can be used if this behavior is desired.) - The implementation should be reasonably performant on all platforms since it uses the asm/word-at-a-time.h API rather than simple byte copy. Kernel-to-kernel string copy is not considered to be performance critical in any case. Signed-off-by: Chris Metcalf Signed-off-by: Greg Kroah-Hartman

lib: memzero_explicit: use barrier instead of OPTIMIZER_HIDE_VAR

2015-05-17T23:12:27+00:00

[ Upstream commit 0b053c9518292705736329a8fe20ef4686ffc8e9 ] OPTIMIZER_HIDE_VAR(), as defined when using gcc, is insufficient to ensure protection from dead store optimization. For the random driver and crypto drivers, calls are emitted ... $ gdb vmlinux (gdb) disassemble memzero_explicit Dump of assembler code for function memzero_explicit: 0xffffffff813a18b0 <+0>: push %rbp 0xffffffff813a18b1 <+1>: mov %rsi,%rdx 0xffffffff813a18b4 <+4>: xor %esi,%esi 0xffffffff813a18b6 <+6>: mov %rsp,%rbp 0xffffffff813a18b9 <+9>: callq 0xffffffff813a7120 0xffffffff813a18be <+14>: pop %rbp 0xffffffff813a18bf <+15>: retq End of assembler dump. (gdb) disassemble extract_entropy [...] 0xffffffff814a5009 <+313>: mov %r12,%rdi 0xffffffff814a500c <+316>: mov $0xa,%esi 0xffffffff814a5011 <+321>: callq 0xffffffff813a18b0 0xffffffff814a5016 <+326>: mov -0x48(%rbp),%rax [...] ... but in case in future we might use facilities such as LTO, then OPTIMIZER_HIDE_VAR() is not sufficient to protect gcc from a possible eviction of the memset(). We have to use a compiler barrier instead. Minimal test example when we assume memzero_explicit() would *not* be a call, but would have been *inlined* instead: static inline void memzero_explicit(void *s, size_t count) { memset(s, 0, count); } int main(void) { char buff[20]; snprintf(buff, sizeof(buff) - 1, "test"); printf("%s", buff); memzero_explicit(buff, sizeof(buff)); return 0; } With := OPTIMIZER_HIDE_VAR(): (gdb) disassemble main Dump of assembler code for function main: [...] 0x0000000000400464 <+36>: callq 0x400410 0x0000000000400469 <+41>: xor %eax,%eax 0x000000000040046b <+43>: add $0x28,%rsp 0x000000000040046f <+47>: retq End of assembler dump. With := barrier(): (gdb) disassemble main Dump of assembler code for function main: [...] 0x0000000000400464 <+36>: callq 0x400410 0x0000000000400469 <+41>: movq $0x0,(%rsp) 0x0000000000400471 <+49>: movq $0x0,0x8(%rsp) 0x000000000040047a <+58>: movl $0x0,0x10(%rsp) 0x0000000000400482 <+66>: xor %eax,%eax 0x0000000000400484 <+68>: add $0x28,%rsp 0x0000000000400488 <+72>: retq End of assembler dump. As can be seen, movq, movq, movl are being emitted inlined via memset(). Reference: http://thread.gmane.org/gmane.linux.kernel.cryptoapi/13764/ Fixes: d4c5efdb9777 ("random: add and use memzero_explicit() for clearing data") Cc: Theodore Ts'o Signed-off-by: mancha security Signed-off-by: Daniel Borkmann Acked-by: Hannes Frederic Sowa Acked-by: Stephan Mueller Signed-off-by: Herbert Xu Signed-off-by: Sasha Levin

Merge tag 'random_for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tytso/random

2014-10-24T19:33:32+00:00

Pull /dev/random updates from Ted Ts'o: "This adds a memzero_explicit() call which is guaranteed not to be optimized away by GCC. This is important when we are wiping cryptographically sensitive material" * tag 'random_for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tytso/random: crypto: memzero_explicit - make sure to clear out sensitive data random: add and use memzero_explicit() for clearing data

random: add and use memzero_explicit() for clearing data

2014-10-17T15:37:29+00:00

zatimend has reported that in his environment (3.16/gcc4.8.3/corei7) memset() calls which clear out sensitive data in extract_{buf,entropy, entropy_user}() in random driver are being optimized away by gcc. Add a helper memzero_explicit() (similarly as explicit_bzero() variants) that can be used in such cases where a variable with sensitive data is being cleared out in the end. Other use cases might also be in crypto code. [ I have put this into lib/string.c though, as it's always built-in and doesn't need any dependencies then. ] Fixes kernel bugzilla: 82041 Reported-by: zatimend@hotmail.co.uk Signed-off-by: Daniel Borkmann Acked-by: Hannes Frederic Sowa Cc: Alexey Dobriyan Signed-off-by: Theodore Ts'o Cc: stable@vger.kernel.org

lib: string: Make all calls to strnicmp into calls to strncasecmp

2014-10-14T00:18:23+00:00

The previous patch made strnicmp into a wrapper for strncasecmp. This patch makes all in-tree users of strnicmp call strncasecmp directly, while still making sure that the strnicmp symbol can be used by out-of-tree modules. It should be considered a temporary hack until all in-tree callers have been converted. Signed-off-by: Rasmus Villemoes Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds

lib/string.c: remove duplicated function

2014-10-14T00:18:23+00:00

lib/string.c contains two functions, strnicmp and strncasecmp, which do roughly the same thing, namely compare two strings case-insensitively up to a given bound. They have slightly different implementations, but the only important difference is that strncasecmp doesn't handle len==0 appropriately; it effectively becomes strcasecmp in that case. strnicmp correctly says that two strings are always equal in their first 0 characters. strncasecmp is the POSIX name for this functionality. So rename the non-broken function to the standard name. To minimize the impact on the rest of the kernel (and since both are exported to modules), make strnicmp a wrapper for strncasecmp. Signed-off-by: Rasmus Villemoes Cc: Grant Likely Cc: Andi Kleen Cc: Dan Carpenter Cc: "H. Peter Anvin" Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds

Make ARCH_HAS_FAST_MULTIPLIER a real config variable

2014-09-13T18:14:53+00:00

It used to be an ad-hoc hack defined by the x86 version of that enabled a couple of library routines to know whether an integer multiply is faster than repeated shifts and additions. This just makes it use the real Kconfig system instead, and makes x86 (which was the only architecture that did this) select the option. NOTE! Even for x86, this really is kind of wrong. If we cared, we would probably not enable this for builds optimized for netburst (P4), where shifts-and-adds are generally faster than multiplies. This patch does *not* change that kind of logic, though, it is purely a syntactic change with no code changes. This was triggered by the fact that we have other places that really want to know "do I want to expand multiples by constants by hand or not", particularly the hash generation code. Signed-off-by: Linus Torvalds

lib/string.c: use the name "C-string" in comments

2014-06-04T23:54:18+00:00

For strncpy() and friends the source string may or may not have an actual NUL character at the end. The documentation is confusing in this because it specifically mentions that you are passing a "NUL-terminated" string. Wikipedia says that "C-string" is an alternative name we can use instead. http://en.wikipedia.org/wiki/Null-terminated_string Signed-off-by: Dan Carpenter Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds