|
From: <sv...@va...> - 2005-04-09 18:25:13
|
Author: sewardj Date: 2005-04-09 19:25:06 +0100 (Sat, 09 Apr 2005) New Revision: 3530 Modified: trunk/NOTES.txt Log: Some notes from memcheck size/endian hackery Modified: trunk/NOTES.txt =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D --- trunk/NOTES.txt 2005-04-09 18:24:19 UTC (rev 3529) +++ trunk/NOTES.txt 2005-04-09 18:25:06 UTC (rev 3530) @@ -1,4 +1,16 @@ =20 +9 Apr 05 (starting work on memcheck for 32/64-bit and big/little endian) +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ +* get rid of memcheck/mc_asm.h and include/tool_asm.h. I think=20 + these are left over from single-platform days, when it made + sense to have tool-helpers written in assembly. Looks like we + need to retain coregrind/core_asm.h, though. + + Urk. Perhaps nuke all that X86_FEAT gunk in coregrind/core_asm.h + though. Vex isn't clever enough to distinguish dozens of CPU + subvariants. + + 23 March 05 ~~~~~~~~~~~ Do we still need ARCH_PTHREQ_RET (or *PTHREQ* for that matter) ? |
|
From: Nicholas N. <nj...@cs...> - 2005-04-09 18:55:44
|
On Sat, 9 Apr 2005 sv...@va... wrote: > +9 Apr 05 (starting work on memcheck for 32/64-bit and big/little endian) > +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ > +* get rid of memcheck/mc_asm.h and include/tool_asm.h. I think=20 > + these are left over from single-platform days, when it made > + sense to have tool-helpers written in assembly. Looks like we > + need to retain coregrind/core_asm.h, though. core_asm.h includes tool_asm.h. The contents of tool_asm.h could be put into core_asm.h, but that would deviate from the structure described at the top of core.h... that structure will slowly change as core.h/tool.h are split up into separate headers for each module, but it might be worthwhile keeping the asm-only headers? Not sure. N |