Details
Diff Detail
Event Timeline
lib/msan/msan_linux.cc | ||
---|---|---|
108–109 | Ah, these calls should be rewritten to check returning values. |
Updated.
- Fixed the #if condition for the case of __mips64.
- SHADOW_TO_ORIGIN() moved out from the platform-specific sections.
- InitShadow replaced with __msan_init in the diagnostic message.
- The calls to TakeMapRange() fixed to check returning values.
- Minor spacing fixes.
lib/msan/msan.h | ||
---|---|---|
66 | It shall be 0x210000000000. Otherwise, the patch works as expected on Linux and FreeBSD. |
lib/msan/msan.h | ||
---|---|---|
66 | Are you saying you will change it to 0x210..? Please do. | |
73 | Please take a look at https://code.google.com/p/memory-sanitizer/issues/detail?id=76 | |
lib/msan/msan_linux.cc | ||
43 | TakeMemoryRange I don't like that it's asymmetrical, i.e. "check that memory is available and maybe protect it". Makes it hard to find a good name for the function. Please either make it "check and either protect or reserve", or just do everything in InitShadow. | |
58 | prot1 .. prot3 are always true, just remove them |
Updated:
- TakeMapRange() is split into two different functions;
- The prot1 and prot2 parameters removed;
- The kShadowSize value fixed;
- A minor fix for the code-in-app-range check.
Please take a look at https://code.google.com/p/memory-sanitizer/issues/detail?id=76
I wonder if it is applicable to FreeBSD as well.
Umm, I suspect we can't really reserve 600000000000-7cffffffffff for Msan needs on FreeBSD. Ed, can you please confirm that?
There are no "traces" in MSan.