Blaisorblade wrote:
From arch/um/os-Linux/sys-i386/registers.c:
/* XXX These need to use [GS]ETFPXREGS and copy_sc_{to,from}_user_skas needs
* to pass in a sufficiently large buffer
*/
Could fixing this be the real fix instead of "fp-state" in the incrementals?
Sorry, no, it could't.
fp-state fixes a wrong handling of fp-/fpx-regs in arch/um/sys-i386/signal.c,
which causes fp-regs of user's "main" routine to be modified, if a
signal-handler
interrupting "main" uses fp.
IMHO, the patch is OK in that it fixes the problem (Please note, Jeff had a
currupted version of the patch in his incrementals for some weeks, that didn't
work and even resulted in worse behavior. The current one is OK.).
What I don't like in the patch, is the duplicated code for fp to fpx conversion
(and vice versa). That code is stolen and modified from
arch/um/sys-i386/ptrace.c
resp. arch/i386/kernel/i387.c. I think, arch/um/sys-i386/ptrace.c should be
reworked to do fp/fpx conversions in memory as signal.c now does. Doing the
rework
ptrace.c also should be extended to support coredumping fp/fpx-regs. All this
were the reasons to call fp-state a "quick and dirty" patch.
Unfortunately I can't do this myself for the moment, because I have to bring up
UML-s390 as fast as possible. So maybe the best thing currently is to use
fp-state.
BTW: having fp-state applied, save_fp_registers and restore_fp_registers in
arch/um/os-Linux/sys-i386/registers.c are no longer needed and might be removed.
Bodo
-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click
_______________________________________________
User-mode-linux-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/user-mode-linux-devel
