|
From: Maynard J. <may...@us...> - 2010-08-09 22:00:08
Attachments:
vg-isa2.05-support.patch
|
Currently, Valgrind has only partial support for new IBM POWER6 instructions (as defined in http://www.power.org/resources/reading/PowerISA_V2.05.pdf). The attached patch completes the 2.05 support. The results of running 'make regtest' on a POWER6 box are almost identical before and after applying this patch -- about 50 failing testcases. I have looked into many of those failing testcases and, so far, have been able to cut the number of failures to less than half that. I have a separate testsuite patch that I will submit separately (upon request, or I'll wait until this patch is accepted) that provides the improved testsuite results. Review comments are welcome. Thanks. -Maynard Signed-off-by: Maynard Johnson <may...@us...> |
|
From: Bart V. A. <bva...@ac...> - 2010-08-10 06:53:51
|
On Mon, Aug 9, 2010 at 11:59 PM, Maynard Johnson <may...@us...>wrote: > Currently, Valgrind has only partial support for new IBM POWER6 > instructions (as defined in > http://www.power.org/resources/reading/PowerISA_V2.05.pdf). The attached > patch completes the 2.05 support. > > The results of running 'make regtest' on a POWER6 box are almost identical > before and after applying this patch -- about 50 failing testcases. I have > looked into many of those failing testcases and, so far, have been able to > cut the number of failures to less than half that. I have a separate > testsuite patch that I will submit separately (upon request, or I'll wait > until this patch is accepted) that provides the improved testsuite results. > > Review comments are welcome. > Which Valgrind version has the patch been developed against ? I have tried to apply the patch to the latest trunk version, but it doesn't seem to apply cleanly: [ ... ] patching file none/tests/ppc64/jm-insns.c Reversed (or previously applied) patch detected! Assume -R? [n] Apply anyway? [n] Skipping patch. 2 out of 2 hunks ignored -- saving rejects to file none/tests/ppc64/jm-insns.c.rej [ ... ] Bart. |
|
From: Maynard J. <may...@us...> - 2010-08-10 15:59:47
|
Bart Van Assche wrote: > On Mon, Aug 9, 2010 at 11:59 PM, Maynard Johnson <may...@us...>wrote: > >> Currently, Valgrind has only partial support for new IBM POWER6 >> instructions (as defined in >> http://www.power.org/resources/reading/PowerISA_V2.05.pdf). The attached >> patch completes the 2.05 support. >> >> The results of running 'make regtest' on a POWER6 box are almost identical >> before and after applying this patch -- about 50 failing testcases. I have >> looked into many of those failing testcases and, so far, have been able to >> cut the number of failures to less than half that. I have a separate >> testsuite patch that I will submit separately (upon request, or I'll wait >> until this patch is accepted) that provides the improved testsuite results. >> >> Review comments are welcome. >> > > Which Valgrind version has the patch been developed against ? > > I have tried to apply the patch to the latest trunk version, but it doesn't > seem to apply cleanly: > > [ ... ] > patching file none/tests/ppc64/jm-insns.c > Reversed (or previously applied) patch detected! Assume -R? [n] > Apply anyway? [n] > Skipping patch. > 2 out of 2 hunks ignored -- saving rejects to file > none/tests/ppc64/jm-insns.c.rej The patch was made against an August 6 SVN snapshot. Sorry, I forgot to mention the rejection of the patch to none/tests/ppc64/jm-insns.c is actually expected. This happens because that file is a symbolic link to none/tests/ppc32/jm-insns.c, which is patched earlier. -Maynard > [ ... ] > > Bart. > |
|
From: Bart V. A. <bva...@ac...> - 2010-08-10 16:31:25
|
On Tue, Aug 10, 2010 at 5:59 PM, Maynard Johnson <may...@us...> wrote: > > Bart Van Assche wrote: > > On Mon, Aug 9, 2010 at 11:59 PM, Maynard Johnson <may...@us...>wrote: > > > >> Currently, Valgrind has only partial support for new IBM POWER6 > >> instructions (as defined in > >> http://www.power.org/resources/reading/PowerISA_V2.05.pdf). The attached > >> patch completes the 2.05 support. > >> > >> The results of running 'make regtest' on a POWER6 box are almost identical > >> before and after applying this patch -- about 50 failing testcases. I have > >> looked into many of those failing testcases and, so far, have been able to > >> cut the number of failures to less than half that. I have a separate > >> testsuite patch that I will submit separately (upon request, or I'll wait > >> until this patch is accepted) that provides the improved testsuite results. > >> > >> Review comments are welcome. > >> > > > > Which Valgrind version has the patch been developed against ? > > > > I have tried to apply the patch to the latest trunk version, but it doesn't > > seem to apply cleanly: > > > > [ ... ] > > patching file none/tests/ppc64/jm-insns.c > > Reversed (or previously applied) patch detected! Assume -R? [n] > > Apply anyway? [n] > > Skipping patch. > > 2 out of 2 hunks ignored -- saving rejects to file > > none/tests/ppc64/jm-insns.c.rej > > The patch was made against an August 6 SVN snapshot. Sorry, I forgot to mention the rejection of the patch to none/tests/ppc64/jm-insns.c is actually expected. This happens because that file is a symbolic link to none/tests/ppc32/jm-insns.c, which is patched earlier. The only other comment I have on the patch is that the indentation style of the VEX changes is different from the surrounding code: the added code has been indented with tabs while the surrounding code is not. Note: since the patch includes VEX changes, only Julian can apply it. Bart. |
|
From: Maynard J. <may...@us...> - 2010-08-10 19:37:19
|
Bart Van Assche wrote: > On Tue, Aug 10, 2010 at 5:59 PM, Maynard Johnson <may...@us...> wrote: >> >> Bart Van Assche wrote: >>> On Mon, Aug 9, 2010 at 11:59 PM, Maynard Johnson <may...@us...>wrote: >>> >>>> Currently, Valgrind has only partial support for new IBM POWER6 >>>> instructions (as defined in >>>> http://www.power.org/resources/reading/PowerISA_V2.05.pdf). The attached >>>> patch completes the 2.05 support. >>>> >>>> The results of running 'make regtest' on a POWER6 box are almost identical >>>> before and after applying this patch -- about 50 failing testcases. I have >>>> looked into many of those failing testcases and, so far, have been able to >>>> cut the number of failures to less than half that. I have a separate >>>> testsuite patch that I will submit separately (upon request, or I'll wait >>>> until this patch is accepted) that provides the improved testsuite results. >>>> >>>> Review comments are welcome. >>>> >>> >>> Which Valgrind version has the patch been developed against ? >>> >>> I have tried to apply the patch to the latest trunk version, but it doesn't >>> seem to apply cleanly: >>> >>> [ ... ] >>> patching file none/tests/ppc64/jm-insns.c >>> Reversed (or previously applied) patch detected! Assume -R? [n] >>> Apply anyway? [n] >>> Skipping patch. >>> 2 out of 2 hunks ignored -- saving rejects to file >>> none/tests/ppc64/jm-insns.c.rej >> >> The patch was made against an August 6 SVN snapshot. Sorry, I forgot to mention the rejection of the patch to none/tests/ppc64/jm-insns.c is actually expected. This happens because that file is a symbolic link to none/tests/ppc32/jm-insns.c, which is patched earlier. > > The only other comment I have on the patch is that the indentation > style of the VEX changes is different from the surrounding code: the > added code has been indented with tabs while the surrounding code is > not. hmmm . . . yeah, ugly. Let me re-swizzle and clean it up a bit. -Maynard > > Note: since the patch includes VEX changes, only Julian can apply it. > > Bart. |
|
From: Maynard J. <may...@us...> - 2010-08-11 15:38:07
Attachments:
vg-isa2.05-support_v2.patch
|
On 08/10/2010 2:36 PM, Maynard Johnson wrote: > Bart Van Assche wrote: >> On Tue, Aug 10, 2010 at 5:59 PM, Maynard Johnson<may...@us...> wrote: >>> >>> Bart Van Assche wrote: >>>> On Mon, Aug 9, 2010 at 11:59 PM, Maynard Johnson<may...@us...>wrote: >>>> >>>>> Currently, Valgrind has only partial support for new IBM POWER6 >>>>> instructions (as defined in >>>>> http://www.power.org/resources/reading/PowerISA_V2.05.pdf). The attached >>>>> patch completes the 2.05 support. >>>>> >>>>> The results of running 'make regtest' on a POWER6 box are almost identical >>>>> before and after applying this patch -- about 50 failing testcases. I have >>>>> looked into many of those failing testcases and, so far, have been able to >>>>> cut the number of failures to less than half that. I have a separate >>>>> testsuite patch that I will submit separately (upon request, or I'll wait >>>>> until this patch is accepted) that provides the improved testsuite results. >>>>> >>>>> Review comments are welcome. >>>>> The attached "version 2" patch fixes the formatting issues that Bart pointed out, plus a couple other minor nits I found while scrubbing it. This patch was made against a snapshot of today's SVN repository. But as I mentioned earlier to Bart, applying the patch will result in an expected failure involving none/tests/ppc64/jm-insns.c, since this file is symbolically linked to none/tests/ppc32/jm-insns.c, which is patched earlier on during the patch application. Just hit "Enter" twice to ignore that failure, and the rest of the patch should apply cleanly. Thanks. -Maynard |
|
From: Maynard J. <may...@us...> - 2010-08-11 20:47:00
|
On 08/11/2010 10:38 AM, Maynard Johnson wrote: > On 08/10/2010 2:36 PM, Maynard Johnson wrote: >> Bart Van Assche wrote: >>> On Tue, Aug 10, 2010 at 5:59 PM, Maynard Johnson<may...@us...> wrote: >>>> >>>> Bart Van Assche wrote: >>>>> On Mon, Aug 9, 2010 at 11:59 PM, Maynard Johnson<may...@us...>wrote: >>>>> >>>>>> Currently, Valgrind has only partial support for new IBM POWER6 >>>>>> instructions (as defined in >>>>>> http://www.power.org/resources/reading/PowerISA_V2.05.pdf). The attached >>>>>> patch completes the 2.05 support. >>>>>> >>>>>> The results of running 'make regtest' on a POWER6 box are almost identical >>>>>> before and after applying this patch -- about 50 failing testcases. I have >>>>>> looked into many of those failing testcases and, so far, have been able to >>>>>> cut the number of failures to less than half that. I have a separate >>>>>> testsuite patch that I will submit separately (upon request, or I'll wait >>>>>> until this patch is accepted) that provides the improved testsuite results. >>>>>> >>>>>> Review comments are welcome. >>>>>> > > The attached "version 2" patch fixes the formatting issues that Bart pointed > out, plus a couple other minor nits I found while scrubbing it. This patch was > made against a snapshot of today's SVN repository. But as I mentioned earlier to > Bart, applying the patch will result in an expected failure involving > none/tests/ppc64/jm-insns.c, since this file is symbolically linked to > none/tests/ppc32/jm-insns.c, which is patched earlier on during the patch > application. Just hit "Enter" twice to ignore that failure, and the rest of the > patch should apply cleanly. Another thing I should have mentioned . . . This patch contains some new shell scripts which need to be changed to executable before attempting to run the testsuite. These new script files are: - memcheck/tests/ppc32/filter_stderr - memcheck/tests/ppc64/filter_stderr - none/tests/ppc32/check_vmx_cap - none/tests/ppc64/check_vmx_cap -Maynard > > Thanks. > -Maynard > > > > ------------------------------------------------------------------------------ > This SF.net email is sponsored by > > Make an app they can't live without > Enter the BlackBerry Developer Challenge > http://p.sf.net/sfu/RIM-dev2dev > > > > _______________________________________________ > Valgrind-developers mailing list > Val...@li... > https://lists.sourceforge.net/lists/listinfo/valgrind-developers |
|
From: Julian S. <js...@ac...> - 2010-08-12 08:41:30
|
Maynard, thanks for the patch. Please could you open a bug report and attach the patches, including the test programs, to the bug report. Patches sent to the mailing list are at (serious) risk of falling through the cracks and disappearing. Also, putting patches in the bug tracker gives a single URL at which we can point anybody interested. See http://www.valgrind.org/support/bug_reports.html for directions on filing bug reports. J On Wednesday, August 11, 2010, Maynard Johnson wrote: > On 08/11/2010 10:38 AM, Maynard Johnson wrote: > > On 08/10/2010 2:36 PM, Maynard Johnson wrote: > >> Bart Van Assche wrote: > >>> On Tue, Aug 10, 2010 at 5:59 PM, Maynard Johnson<may...@us...> wrote: > >>>> Bart Van Assche wrote: > >>>>> On Mon, Aug 9, 2010 at 11:59 PM, Maynard Johnson<may...@us...>wrote: > >>>>>> Currently, Valgrind has only partial support for new IBM POWER6 > >>>>>> instructions (as defined in > >>>>>> http://www.power.org/resources/reading/PowerISA_V2.05.pdf). The > >>>>>> attached patch completes the 2.05 support. > >>>>>> > >>>>>> The results of running 'make regtest' on a POWER6 box are almost > >>>>>> identical before and after applying this patch -- about 50 failing > >>>>>> testcases. I have looked into many of those failing testcases and, > >>>>>> so far, have been able to cut the number of failures to less than > >>>>>> half that. I have a separate testsuite patch that I will submit > >>>>>> separately (upon request, or I'll wait until this patch is > >>>>>> accepted) that provides the improved testsuite results. > >>>>>> > >>>>>> Review comments are welcome. > > > > The attached "version 2" patch fixes the formatting issues that Bart > > pointed out, plus a couple other minor nits I found while scrubbing it. > > This patch was made against a snapshot of today's SVN repository. But as > > I mentioned earlier to Bart, applying the patch will result in an > > expected failure involving none/tests/ppc64/jm-insns.c, since this file > > is symbolically linked to none/tests/ppc32/jm-insns.c, which is patched > > earlier on during the patch application. Just hit "Enter" twice to > > ignore that failure, and the rest of the patch should apply cleanly. > > Another thing I should have mentioned . . . This patch contains some new > shell scripts which need to be changed to executable before attempting to > run the testsuite. These new script files are: > - memcheck/tests/ppc32/filter_stderr > - memcheck/tests/ppc64/filter_stderr > - none/tests/ppc32/check_vmx_cap > - none/tests/ppc64/check_vmx_cap > > -Maynard > > > Thanks. > > -Maynard > > > > > > > > ------------------------------------------------------------------------- > > ----- This SF.net email is sponsored by > > > > Make an app they can't live without > > Enter the BlackBerry Developer Challenge > > http://p.sf.net/sfu/RIM-dev2dev > > > > > > > > _______________________________________________ > > Valgrind-developers mailing list > > Val...@li... > > https://lists.sourceforge.net/lists/listinfo/valgrind-developers > > --------------------------------------------------------------------------- > --- This SF.net email is sponsored by > > Make an app they can't live without > Enter the BlackBerry Developer Challenge > http://p.sf.net/sfu/RIM-dev2dev > _______________________________________________ > Valgrind-developers mailing list > Val...@li... > https://lists.sourceforge.net/lists/listinfo/valgrind-developers |
|
From: Maynard J. <may...@us...> - 2010-08-12 14:35:40
|
On 08/12/2010 3:44 AM, Julian Seward wrote: > > Maynard, thanks for the patch. > > Please could you open a bug report and attach the patches, including > the test programs, to the bug report. Patches sent to the mailing list > are at (serious) risk of falling through the cracks and disappearing. > Also, putting patches in the bug tracker gives a single URL at which > we can point anybody interested. Julian, Thanks for the reply. I opened a bug report as you asked: https://bugs.kde.org/show_bug.cgi?id=247526. -Maynard > > See http://www.valgrind.org/support/bug_reports.html for directions > on filing bug reports. > > J > > On Wednesday, August 11, 2010, Maynard Johnson wrote: >> On 08/11/2010 10:38 AM, Maynard Johnson wrote: >>> On 08/10/2010 2:36 PM, Maynard Johnson wrote: >>>> Bart Van Assche wrote: >>>>> On Tue, Aug 10, 2010 at 5:59 PM, Maynard Johnson<may...@us...> > wrote: >>>>>> Bart Van Assche wrote: >>>>>>> On Mon, Aug 9, 2010 at 11:59 PM, Maynard > Johnson<may...@us...>wrote: >>>>>>>> Currently, Valgrind has only partial support for new IBM POWER6 >>>>>>>> instructions (as defined in >>>>>>>> http://www.power.org/resources/reading/PowerISA_V2.05.pdf). The >>>>>>>> attached patch completes the 2.05 support. >>>>>>>> >>>>>>>> The results of running 'make regtest' on a POWER6 box are almost >>>>>>>> identical before and after applying this patch -- about 50 failing >>>>>>>> testcases. I have looked into many of those failing testcases and, >>>>>>>> so far, have been able to cut the number of failures to less than >>>>>>>> half that. I have a separate testsuite patch that I will submit >>>>>>>> separately (upon request, or I'll wait until this patch is >>>>>>>> accepted) that provides the improved testsuite results. >>>>>>>> >>>>>>>> Review comments are welcome. >>> >>> The attached "version 2" patch fixes the formatting issues that Bart >>> pointed out, plus a couple other minor nits I found while scrubbing it. >>> This patch was made against a snapshot of today's SVN repository. But as >>> I mentioned earlier to Bart, applying the patch will result in an >>> expected failure involving none/tests/ppc64/jm-insns.c, since this file >>> is symbolically linked to none/tests/ppc32/jm-insns.c, which is patched >>> earlier on during the patch application. Just hit "Enter" twice to >>> ignore that failure, and the rest of the patch should apply cleanly. >> >> Another thing I should have mentioned . . . This patch contains some new >> shell scripts which need to be changed to executable before attempting to >> run the testsuite. These new script files are: >> - memcheck/tests/ppc32/filter_stderr >> - memcheck/tests/ppc64/filter_stderr >> - none/tests/ppc32/check_vmx_cap >> - none/tests/ppc64/check_vmx_cap >> >> -Maynard >> >>> Thanks. >>> -Maynard >>> >>> >>> >>> ------------------------------------------------------------------------- >>> ----- This SF.net email is sponsored by >>> >>> Make an app they can't live without >>> Enter the BlackBerry Developer Challenge >>> http://p.sf.net/sfu/RIM-dev2dev >>> >>> >>> >>> _______________________________________________ >>> Valgrind-developers mailing list >>> Val...@li... >>> https://lists.sourceforge.net/lists/listinfo/valgrind-developers >> >> --------------------------------------------------------------------------- >> --- This SF.net email is sponsored by >> >> Make an app they can't live without >> Enter the BlackBerry Developer Challenge >> http://p.sf.net/sfu/RIM-dev2dev >> _______________________________________________ >> Valgrind-developers mailing list >> Val...@li... >> https://lists.sourceforge.net/lists/listinfo/valgrind-developers > |