|
From: Carl L. <ce...@us...> - 2018-07-10 16:29:45
|
Julian:
Just wondering if you had picked a date for the next release of
Valgrind? It has been a little over a year now since the last release.
I am starting to get a few requests for patches/backports to the last
release for the various PPC fixes that are in mainline. It would be
nice to point people to a new version rather then a handful of patches
for just PPC.
Anyway, just thought I would see where you are at in terms of the next
release. Thanks.
Carl Love
|
|
From: Mark W. <ma...@kl...> - 2018-07-11 08:59:14
|
Hi, On Tue, 2018-07-10 at 09:19 -0700, Carl Love wrote: > Just wondering if you had picked a date for the next release of > Valgrind? It has been a little over a year now since the last release. > I am starting to get a few requests for patches/backports to the last > release for the various PPC fixes that are in mainline. It would be > nice to point people to a new version rather then a handful of patches > for just PPC. I would also like to see a new release. For Fedora I am currently carrying 22 patches for various kernel, gcc, glibc and binutils updates. It would be good to get back to "pure upstream" again. Before doing a release it would be good to at least try to see if we can resolve the issue with newer binutils and ld.so being linked with ld -z separate-code as seen with Debian: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=903389 https://sourceware.org/bugzilla/show_bug.cgi?id=23357 https://bugs.kde.org/show_bug.cgi?id=395682 But I just saw a report that it also happens on Fedora rawhide with latest binutils/glibc. Cheers, Mark P.S. The current list of backports in the fedora valgrind.spec are: https://src.fedoraproject.org/cgit/rpms/valgrind.git/tree/ # KDE#381272 ppc64 doesn't compile test_isa_2_06_partx.c without VSX support # KDE#381289 epoll_pwait can have a NULL sigmask. # KDE#381274 powerpc too chatty even with --sigill-diagnostics=no # KDE#381556 arm64: Handle feature registers access on 4.11 Linux kernel # Workaround that masks CPUID support in HWCAP on aarch64 (#1464211) # RHBZ#1466017 ARM ld.so index warnings. # KDE#381805 arm32 needs ld.so index hardwire for new glibc security fixes # KDE#381769 Use ucontext_t instead of struct ucontext # valgrind svn r16453 Fix some tests failure with GDB 8.0 # valgrind svn r16454. disable vgdb poll in the child after fork # KDE#382998 xml-socket doesn't work # KDE#385334 # PPC64, vpermr, xxperm, xxpermr fix Iop_Perm8x16 selector field # PPC64, revert the change to vperm instruction. # KDE#385183 # PPC64, Add support for xscmpeqdp, xscmpgtdp, xscmpgedp, xsmincdp instructions # PPC64, Fix bug in vperm instruction. # KDE#385210 # PPC64, Re-implement the vpermr instruction using the Iop_Perm8x16. # KDE#385208 # PPC64, Use the vperm code to implement the xxperm inst. # PPC64, Replace body of generate_store_FPRF with C helper function. # PPC64, Add support for the Data Stream Control Register (DSCR) # Fix eflags handling in amd64 instruction tests # KDE#385868 ld.so _dl_runtime_resolve_avx_slow conditional jump warning # Implement static TLS code for more platforms # KDE#386397 PPC64 valgrind truncates powerpc timebase to 32-bits. # KDE#387773 Files in .gnu_debugaltlink should be resolved relative to .debug # KDE#387712 s390x cgijnl reports Conditional jump depends on uninit value # KDE#391164 constraint bug in tests/ppc64/test_isa_2_07_part1.c for mtfprwa # KDE#393062 Reading build-id ELF note "debuginfo reader: ensure_valid failed" # KDE#368913 WARNING: unhandled arm64-linux syscall: 117 (ptrace) |
|
From: Philippe W. <phi...@sk...> - 2018-07-12 22:21:43
|
+1 to have a release one of these weeks. I can do the usual 'run valgrind under valgrind' to detect possible leaks or other bugs inside valgrind. Note that there are a bunch of bugs which seems good to fix (e.g. the s390x https://bugs.kde.org/show_bug.cgi?id=385409 and https://bugs.kde.org/show_bug.cgi?id=385412 IIRC, I quickly looked at them some time ago, and they were only touching s390x related logic). Would be nice to have a list of 'important to fix' bugs (e.g. https://bugs.kde.org/show_bug.cgi?id=395682 Reading debug info of binaries with readonly PT_LOAD segments) but maybe others ? Philippe On Wed, 2018-07-11 at 10:59 +0200, Mark Wielaard wrote: > Hi, > > On Tue, 2018-07-10 at 09:19 -0700, Carl Love wrote: > > Just wondering if you had picked a date for the next release of > > Valgrind? It has been a little over a year now since the last release. > > I am starting to get a few requests for patches/backports to the last > > release for the various PPC fixes that are in mainline. It would be > > nice to point people to a new version rather then a handful of patches > > for just PPC. > > I would also like to see a new release. For Fedora I am currently > carrying 22 patches for various kernel, gcc, glibc and binutils updates. > It would be good to get back to "pure upstream" again. > > Before doing a release it would be good to at least try to see if we > can resolve the issue with newer binutils and ld.so being linked with > ld -z separate-code as seen with Debian: > https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=903389 > https://sourceware.org/bugzilla/show_bug.cgi?id=23357 > https://bugs.kde.org/show_bug.cgi?id=395682 > But I just saw a report that it also happens on Fedora rawhide with > latest binutils/glibc. > > Cheers, > > Mark > > P.S. The current list of backports in the fedora valgrind.spec are: > https://src.fedoraproject.org/cgit/rpms/valgrind.git/tree/ > > # KDE#381272 ppc64 doesn't compile test_isa_2_06_partx.c without VSX support > > # KDE#381289 epoll_pwait can have a NULL sigmask. > > # KDE#381274 powerpc too chatty even with --sigill-diagnostics=no > > # KDE#381556 arm64: Handle feature registers access on 4.11 Linux kernel > # Workaround that masks CPUID support in HWCAP on aarch64 (#1464211) > > # RHBZ#1466017 ARM ld.so index warnings. > # KDE#381805 arm32 needs ld.so index hardwire for new glibc security fixes > > # KDE#381769 Use ucontext_t instead of struct ucontext > > # valgrind svn r16453 Fix some tests failure with GDB 8.0 > > # valgrind svn r16454. disable vgdb poll in the child after fork > > # KDE#382998 xml-socket doesn't work > > # KDE#385334 > # PPC64, vpermr, xxperm, xxpermr fix Iop_Perm8x16 selector field > # PPC64, revert the change to vperm instruction. > # KDE#385183 > # PPC64, Add support for xscmpeqdp, xscmpgtdp, xscmpgedp, xsmincdp instructions > # PPC64, Fix bug in vperm instruction. > # KDE#385210 > # PPC64, Re-implement the vpermr instruction using the Iop_Perm8x16. > # KDE#385208 > # PPC64, Use the vperm code to implement the xxperm inst. > # PPC64, Replace body of generate_store_FPRF with C helper function. > # PPC64, Add support for the Data Stream Control Register (DSCR) > > # Fix eflags handling in amd64 instruction tests > > # KDE#385868 ld.so _dl_runtime_resolve_avx_slow conditional jump warning > > # Implement static TLS code for more platforms > > # KDE#386397 PPC64 valgrind truncates powerpc timebase to 32-bits. > > # KDE#387773 Files in .gnu_debugaltlink should be resolved relative to .debug > > # KDE#387712 s390x cgijnl reports Conditional jump depends on uninit value > > # KDE#391164 constraint bug in tests/ppc64/test_isa_2_07_part1.c for mtfprwa > > # KDE#393062 Reading build-id ELF note "debuginfo reader: ensure_valid failed" > > # KDE#368913 WARNING: unhandled arm64-linux syscall: 117 (ptrace) > > ------------------------------------------------------------------------------ > Check out the vibrant tech community on one of the world's most > engaging tech sites, Slashdot.org! http://sdm.link/slashdot > _______________________________________________ > Valgrind-developers mailing list > Val...@li... > https://lists.sourceforge.net/lists/listinfo/valgrind-developers |
|
From: Christian B. <bor...@de...> - 2018-07-13 08:51:50
|
On 07/13/2018 12:21 AM, Philippe Waroquiers wrote: > +1 to have a release one of these weeks. +1 We can certainly test this on s390x and give feedback. Would be good to have z13 patches integrated though. > > I can do the usual 'run valgrind under valgrind' to detect > possible leaks or other bugs inside valgrind. > > Note that there are a bunch of bugs which seems good to fix > (e.g. the s390x https://bugs.kde.org/show_bug.cgi?id=385409 > and https://bugs.kde.org/show_bug.cgi?id=385412 > IIRC, I quickly looked at them some time ago, and they > were only touching s390x related logic). > > Would be nice to have a list of 'important to fix' bugs > (e.g. https://bugs.kde.org/show_bug.cgi?id=395682 > Reading debug info of binaries with readonly PT_LOAD segments) > but maybe others ? > > Philippe > > > On Wed, 2018-07-11 at 10:59 +0200, Mark Wielaard wrote: >> Hi, >> >> On Tue, 2018-07-10 at 09:19 -0700, Carl Love wrote: >>> Just wondering if you had picked a date for the next release of >>> Valgrind? It has been a little over a year now since the last release. >>> I am starting to get a few requests for patches/backports to the last >>> release for the various PPC fixes that are in mainline. It would be >>> nice to point people to a new version rather then a handful of patches >>> for just PPC. >> >> I would also like to see a new release. For Fedora I am currently >> carrying 22 patches for various kernel, gcc, glibc and binutils updates. >> It would be good to get back to "pure upstream" again. >> >> Before doing a release it would be good to at least try to see if we >> can resolve the issue with newer binutils and ld.so being linked with >> ld -z separate-code as seen with Debian: >> https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=903389 >> https://sourceware.org/bugzilla/show_bug.cgi?id=23357 >> https://bugs.kde.org/show_bug.cgi?id=395682 >> But I just saw a report that it also happens on Fedora rawhide with >> latest binutils/glibc. >> >> Cheers, >> >> Mark >> >> P.S. The current list of backports in the fedora valgrind.spec are: >> https://src.fedoraproject.org/cgit/rpms/valgrind.git/tree/ >> >> # KDE#381272 ppc64 doesn't compile test_isa_2_06_partx.c without VSX support >> >> # KDE#381289 epoll_pwait can have a NULL sigmask. >> >> # KDE#381274 powerpc too chatty even with --sigill-diagnostics=no >> >> # KDE#381556 arm64: Handle feature registers access on 4.11 Linux kernel >> # Workaround that masks CPUID support in HWCAP on aarch64 (#1464211) >> >> # RHBZ#1466017 ARM ld.so index warnings. >> # KDE#381805 arm32 needs ld.so index hardwire for new glibc security fixes >> >> # KDE#381769 Use ucontext_t instead of struct ucontext >> >> # valgrind svn r16453 Fix some tests failure with GDB 8.0 >> >> # valgrind svn r16454. disable vgdb poll in the child after fork >> >> # KDE#382998 xml-socket doesn't work >> >> # KDE#385334 >> # PPC64, vpermr, xxperm, xxpermr fix Iop_Perm8x16 selector field >> # PPC64, revert the change to vperm instruction. >> # KDE#385183 >> # PPC64, Add support for xscmpeqdp, xscmpgtdp, xscmpgedp, xsmincdp instructions >> # PPC64, Fix bug in vperm instruction. >> # KDE#385210 >> # PPC64, Re-implement the vpermr instruction using the Iop_Perm8x16. >> # KDE#385208 >> # PPC64, Use the vperm code to implement the xxperm inst. >> # PPC64, Replace body of generate_store_FPRF with C helper function. >> # PPC64, Add support for the Data Stream Control Register (DSCR) >> >> # Fix eflags handling in amd64 instruction tests >> >> # KDE#385868 ld.so _dl_runtime_resolve_avx_slow conditional jump warning >> >> # Implement static TLS code for more platforms >> >> # KDE#386397 PPC64 valgrind truncates powerpc timebase to 32-bits. >> >> # KDE#387773 Files in .gnu_debugaltlink should be resolved relative to .debug >> >> # KDE#387712 s390x cgijnl reports Conditional jump depends on uninit value >> >> # KDE#391164 constraint bug in tests/ppc64/test_isa_2_07_part1.c for mtfprwa >> >> # KDE#393062 Reading build-id ELF note "debuginfo reader: ensure_valid failed" >> >> # KDE#368913 WARNING: unhandled arm64-linux syscall: 117 (ptrace) >> >> ------------------------------------------------------------------------------ >> Check out the vibrant tech community on one of the world's most >> engaging tech sites, Slashdot.org! http://sdm.link/slashdot >> _______________________________________________ >> Valgrind-developers mailing list >> Val...@li... >> https://lists.sourceforge.net/lists/listinfo/valgrind-developers > > ------------------------------------------------------------------------------ > Check out the vibrant tech community on one of the world's most > engaging tech sites, Slashdot.org! http://sdm.link/slashdot > _______________________________________________ > Valgrind-developers mailing list > Val...@li... > https://lists.sourceforge.net/lists/listinfo/valgrind-developers > |
|
From: Julian S. <js...@ac...> - 2018-07-13 08:12:21
|
Hi Carl, [All], I propose the following: * Up to 23 Aug: fix as many bugs as we can * Fri 23 Aug: code freeze. After this, only "important" bug fixes to land * Mon 10 Sept: release At an absolute minimum I would like to have the s390/z13 fixes in, and the current PT_LOAD problem(s) resolved by then. Plus of course as many other fixes as we have time for. I will create a docs/internals/3_13_BUGSTATUS.txt file in the next day or so, so as to get a top-level overview of the bug situation. J On 10/07/18 18:19, Carl Love wrote: > Julian: > > Just wondering if you had picked a date for the next release of > Valgrind? It has been a little over a year now since the last release. > I am starting to get a few requests for patches/backports to the last > release for the various PPC fixes that are in mainline. It would be > nice to point people to a new version rather then a handful of patches > for just PPC. > > Anyway, just thought I would see where you are at in terms of the next > release. Thanks. > > Carl Love > > > ------------------------------------------------------------------------------ > Check out the vibrant tech community on one of the world's most > engaging tech sites, Slashdot.org! http://sdm.link/slashdot > _______________________________________________ > Valgrind-developers mailing list > Val...@li... > https://lists.sourceforge.net/lists/listinfo/valgrind-developers > |
|
From: Mark W. <ma...@kl...> - 2018-07-16 13:32:38
|
On Fri, 2018-07-13 at 09:55 +0200, Julian Seward wrote: > I propose the following: > > * Up to 23 Aug: fix as many bugs as we can > * Fri 23 Aug: code freeze. After this, only "important" bug fixes to > land > * Mon 10 Sept: release That would nicely line up with the Fedora 29 schedule: https://fedoraproject.org/wiki/Releases/29/Schedule Beta Freeze: 2018-08-28 Final Freeze: 2018-10-09 It would be slightly convenient to have the final release on Friday 7 Sept, then it will be easier to slip it in before the final freeze. Note that The GNU Tools Cauldron is from September 7 to 9 in Manchester, UK. https://gcc.gnu.org/wiki/cauldron2018 I will attend, maybe others would like to join there (we could proudly show off our new release :) I will be away from the interwebs from 13 Augustus till 28 Augustus, so cannot help with the code freeze event. But should be back in time to help with the final release. If helpful, I could help with doing a pre-release (alpha) build on Friday August 10 (then we could call the Friday 23 Aug one the beta). > At an absolute minimum I would like to have the s390/z13 fixes in, > and the current PT_LOAD problem(s) resolved by then. I did just commit the fix for bug #395682. But that only fixes that particular issue (ld -zseparate-code) . There are at least two other mmap/segment/section load/mapping issues: https://bugs.kde.org/show_bug.cgi?id=396476 (zero-size PT_LOAD segments) https://bugs.kde.org/show_bug.cgi?id=390871 (multiple .rodata* sections) It would also be good to get some testing with alternative linkers and/or on MacOS and Solaris. Sadly this code is somewhat fragile. Cheers, Mark |
|
From: Philippe W. <phi...@sk...> - 2018-09-02 13:55:15
|
On Fri, 2018-07-13 at 09:55 +0200, Julian Seward wrote: > Hi Carl, [All], > > I propose the following: > > * Up to 23 Aug: fix as many bugs as we can > * Fri 23 Aug: code freeze. After this, only "important" bug fixes to land > * Mon 10 Sept: release > > At an absolute minimum I would like to have the s390/z13 fixes in, and the > current PT_LOAD problem(s) resolved by then. Plus of course as many other > fixes as we have time for. I will create a docs/internals/3_13_BUGSTATUS.txt > file in the next day or so, so as to get a top-level overview of the bug > situation. Hello all, Recently, I have done another change for 393146 - failing assert "is_DebugInfo_active(di)" (at least, this solves the case reproduced on gnu/linux with qt5). I also updated gdbserver tests for glibc 2.27. I do not think I have anything more burning waiting for a fix in the 3.14 release. Is there anything needed for 3.14 still blocking/not done yet? (the s390/z13 fixes maybe ?) Thanks Philippe |
|
From: Mark W. <ma...@kl...> - 2018-09-17 21:31:59
|
Hi, On Sun, Sep 02, 2018 at 03:55:00PM +0200, Philippe Waroquiers wrote: > Recently, I have done another change for > 393146 - failing assert "is_DebugInfo_active(di)" > (at least, this solves the case reproduced on gnu/linux with qt5). > I also updated gdbserver tests for glibc 2.27. > > I do not think I have anything more burning waiting for a fix in > the 3.14 release. I also don't have anything urgent. > Is there anything needed for 3.14 still blocking/not done yet? > (the s390/z13 fixes maybe ?) I think everybody was a bit too busy with other things. But I do think current GIT is ready for release. It doesn't have all the s390x/z13/trapping instruction support yet though. It would be really good to do a release soon. With distros updating to newer GCC, binutils, glibc, etc. they really need a new release. And as far as I can see it is better, without regressions, compared the 3.13.0. For Fedora I took current GIT (commit 5d41dade) and build for rawhide: https://koji.fedoraproject.org/koji/buildinfo?buildID=1144750 There are also binaries for CentOS6 (i386, x86_64), CentOS7 (x86_64, ppc64le) and Fedora 28 (i386, x86_64, ppc64le) on copr: https://copr.fedorainfracloud.org/coprs/mjw/valgrind-3.14.0/ The test results look pretty reasonable. Cheers, Mark |
|
From: Philippe W. <phi...@sk...> - 2018-09-17 21:50:58
|
On Mon, 2018-09-17 at 23:31 +0200, Mark Wielaard wrote: > Hi, > > On Sun, Sep 02, 2018 at 03:55:00PM +0200, Philippe Waroquiers wrote: > > Recently, I have done another change for > > 393146 - failing assert "is_DebugInfo_active(di)" > > (at least, this solves the case reproduced on gnu/linux with qt5). > > I also updated gdbserver tests for glibc 2.27. > > > > I do not think I have anything more burning waiting for a fix in > > the 3.14 release. > > I also don't have anything urgent. Nothing new for me, and I used a recent 3.14-GIT at work without any problems. So, ok to release from my selfish point of view :). The bug : Bug 398028 - Assertion `csfi_fits` failing in simple C program with embedded Julia code. has however been reported also by another user on valgrind-users. So, there is a problem in this area (maybe triggered by a newer gcc or glibc or ...). Philippe |
|
From: Rhys K. <rhy...@gm...> - 2018-09-18 04:06:01
|
On Mon, 17 Sep 2018 at 17:51, Philippe Waroquiers < phi...@sk...> wrote: > On Mon, 2018-09-17 at 23:31 +0200, Mark Wielaard wrote: > > Hi, > > > > On Sun, Sep 02, 2018 at 03:55:00PM +0200, Philippe Waroquiers wrote: > > > Recently, I have done another change for > > > 393146 - failing assert "is_DebugInfo_active(di)" > > > (at least, this solves the case reproduced on gnu/linux with qt5). > > > I also updated gdbserver tests for glibc 2.27. > > > > > > I do not think I have anything more burning waiting for a fix in > > > the 3.14 release. > > > > I also don't have anything urgent. > Nothing new for me, and I used a recent 3.14-GIT at work > without any problems. > So, ok to release from my selfish point of view :). > I'm okay for a release 3.14 from the perspective of macOS. We've got some important bug fixes for this platform in git, which it would be great to have a packaged release out there which includes them. > > The bug : > Bug 398028 - Assertion `csfi_fits` failing in simple C program with > embedded Julia code. > has however been reported also by another user on valgrind-users. > So, there is a problem in this area (maybe triggered by a newer gcc or > glibc or ...). > > Philippe > > > > _______________________________________________ > Valgrind-developers mailing list > Val...@li... > https://lists.sourceforge.net/lists/listinfo/valgrind-developers > |