| Message ID | 20210804172259.9143-1-petr.vorel@gmail.com |
|---|---|
| State | Changes Requested |
| Headers | show |
| Series | [RFC,1/1] package/glibc: bump version to 2.34-5-g9995d0588f4f9adc68419224d2b3698e2ca4f77e | expand |
Hello Petr, On Wed, 4 Aug 2021 19:22:59 +0200 Petr Vorel <petr.vorel@gmail.com> wrote: > Also upgrade to on RISC-V (that could have been done in d5ce3b5db5) > > Signed-off-by: Petr Vorel <petr.vorel@gmail.com> > --- > Hi, > > I tested it with building pc_x86_64_efi_defconfig, adding nfs-utils, > ltp-testsuite and few other tools. Not sure if this is enough for > testing. I think that what Romain Naour typically does is run all qemu_* defconfigs (of course tweaked to use glibc instead of uclibc), and push them to Gitlab, so that they get build tested *and* boot tested under Qemu. Thomas
On Thu, Aug 5, 2021 at 3:23 AM Petr Vorel <petr.vorel@gmail.com> wrote: > > Also upgrade to on RISC-V (that could have been done in d5ce3b5db5) > > Signed-off-by: Petr Vorel <petr.vorel@gmail.com> > --- > Hi, > > I tested it with building pc_x86_64_efi_defconfig, adding nfs-utils, > ltp-testsuite and few other tools. Not sure if this is enough for > testing. > > Kind regards, > Petr > > .../glibc.hash | 7 ------- > .../glibc.hash | 2 +- > package/glibc/glibc.mk | 8 +------- > 3 files changed, 2 insertions(+), 15 deletions(-) > delete mode 100644 package/glibc/2.32.9000-69-gbd394d131c10c9ec22c6424197b79410042eed99/glibc.hash > rename package/glibc/{2.33-46-gedfd11197ecf3629bbb4b66c5814da09a61a7f9f => 2.34-5-g9995d0588f4f9adc68419224d2b3698e2ca4f77e}/glibc.hash (70%) > > diff --git a/package/glibc/2.32.9000-69-gbd394d131c10c9ec22c6424197b79410042eed99/glibc.hash b/package/glibc/2.32.9000-69-gbd394d131c10c9ec22c6424197b79410042eed99/glibc.hash > deleted file mode 100644 > index 558591522e..0000000000 > --- a/package/glibc/2.32.9000-69-gbd394d131c10c9ec22c6424197b79410042eed99/glibc.hash > +++ /dev/null > @@ -1,7 +0,0 @@ > -# Locally calculated (fetched from Github) > -sha256 c3828224d69a42f5b6982cbec8ae01cf494fe975ce9023e36ff2c10a2fb93e42 glibc-2.32.9000-69-gbd394d131c10c9ec22c6424197b79410042eed99.tar.gz > - > -# Hashes for license files > -sha256 8177f97513213526df2cf6184d8ff986c675afb514d4e68a404010521b880643 COPYING > -sha256 dc626520dcd53a22f727af3ee42c770e56c97a64fe3adb063799d8ab032fe551 COPYING.LIB > -sha256 b33d0bd9f685b46853548814893a6135e74430d12f6d94ab3eba42fc591f83bc LICENSES > diff --git a/package/glibc/2.33-46-gedfd11197ecf3629bbb4b66c5814da09a61a7f9f/glibc.hash b/package/glibc/2.34-5-g9995d0588f4f9adc68419224d2b3698e2ca4f77e/glibc.hash > similarity index 70% > rename from package/glibc/2.33-46-gedfd11197ecf3629bbb4b66c5814da09a61a7f9f/glibc.hash > rename to package/glibc/2.34-5-g9995d0588f4f9adc68419224d2b3698e2ca4f77e/glibc.hash > index 4b8d8297ec..8b5aacb535 100644 > --- a/package/glibc/2.33-46-gedfd11197ecf3629bbb4b66c5814da09a61a7f9f/glibc.hash > +++ b/package/glibc/2.34-5-g9995d0588f4f9adc68419224d2b3698e2ca4f77e/glibc.hash > @@ -1,5 +1,5 @@ > # Locally calculated (fetched from Github) > -sha256 3e2004b16ec6b79a19a88ed4736719ef9e419f9d8e7013faced654c0339e23dc glibc-2.33-46-gedfd11197ecf3629bbb4b66c5814da09a61a7f9f.tar.gz > +sha256 77af51b251591d013ad4634863a42ac8ce62f4b2c7c29cba38a2ad5098d2727f glibc-2.34-5-g9995d0588f4f9adc68419224d2b3698e2ca4f77e.tar.gz > > # Hashes for license files > sha256 8177f97513213526df2cf6184d8ff986c675afb514d4e68a404010521b880643 COPYING > diff --git a/package/glibc/glibc.mk b/package/glibc/glibc.mk > index 1c8c409878..320848bd01 100644 > --- a/package/glibc/glibc.mk > +++ b/package/glibc/glibc.mk > @@ -11,13 +11,7 @@ else > # Generate version string using: > # git describe --match 'glibc-*' --abbrev=40 origin/release/MAJOR.MINOR/master | cut -d '-' -f 2- > # When updating the version, please also update localedef > -ifeq ($(BR2_RISCV_32),y) > -# RISC-V 32-bit (RV32) requires glibc 2.33 or newer > -# Until 2.33 is released, just use master > -GLIBC_VERSION = 2.32.9000-69-gbd394d131c10c9ec22c6424197b79410042eed99 > -else > -GLIBC_VERSION = 2.33-46-gedfd11197ecf3629bbb4b66c5814da09a61a7f9f > -endif > +GLIBC_VERSION = 2.34-5-g9995d0588f4f9adc68419224d2b3698e2ca4f77e Woo! For the RISC-V part Reviewed-by: Alistair Francis <alistair.francis@wdc.com> Alistair > # Upstream doesn't officially provide an https download link. > # There is one (https://sourceware.org/git/glibc.git) but it's not reliable, > # sometimes the connection times out. So use an unofficial github mirror. > -- > 2.32.0 > > _______________________________________________ > buildroot mailing list > buildroot@busybox.net > http://lists.busybox.net/mailman/listinfo/buildroot
Hello, Le 04/08/2021 à 21:30, Thomas Petazzoni a écrit : > Hello Petr, > > On Wed, 4 Aug 2021 19:22:59 +0200 > Petr Vorel <petr.vorel@gmail.com> wrote: > >> Also upgrade to on RISC-V (that could have been done in d5ce3b5db5) >> >> Signed-off-by: Petr Vorel <petr.vorel@gmail.com> >> --- >> Hi, >> >> I tested it with building pc_x86_64_efi_defconfig, adding nfs-utils, >> ltp-testsuite and few other tools. Not sure if this is enough for >> testing. > > I think that what Romain Naour typically does is run all qemu_* > defconfigs (of course tweaked to use glibc instead of uclibc), and push > them to Gitlab, so that they get build tested *and* boot tested under > Qemu. Indeed, the runtime test is only a boot test. I've looked at the glibc testsuite but it's difficult to setup with a cross-compiled system running under Qemu in a CI (glitlab). For testing, upstream suggest to use a NFS to share the glibc sources between the host and the target (not very handy): https://sourceware.org/glibc/wiki/Testing/Testsuite Some tests requires a minimal amount of RAM that is not always provided by Qemu target. There is also libc-test project from musl project: https://wiki.musl-libc.org/libc-test.html Indeed running some tests from ltp-testsuite would be better than nothing. On the other hand we can detect enough boot issue related to toolchain upgrade to keep us busy. Best regards, Romain > > Thomas >
Hi Romain, > Hello, > Le 04/08/2021 à 21:30, Thomas Petazzoni a écrit : > > Hello Petr, > > On Wed, 4 Aug 2021 19:22:59 +0200 > > Petr Vorel <petr.vorel@gmail.com> wrote: > >> Also upgrade to on RISC-V (that could have been done in d5ce3b5db5) > >> Signed-off-by: Petr Vorel <petr.vorel@gmail.com> > >> --- > >> Hi, > >> I tested it with building pc_x86_64_efi_defconfig, adding nfs-utils, > >> ltp-testsuite and few other tools. Not sure if this is enough for > >> testing. > > I think that what Romain Naour typically does is run all qemu_* > > defconfigs (of course tweaked to use glibc instead of uclibc), and push > > them to Gitlab, so that they get build tested *and* boot tested under > > Qemu. > Indeed, the runtime test is only a boot test. I've looked at the glibc testsuite > but it's difficult to setup with a cross-compiled system running under Qemu in a > CI (glitlab). I hoped there would be some setup ready, but understand there is nothing. > For testing, upstream suggest to use a NFS to share the glibc sources between > the host and the target (not very handy): > https://sourceware.org/glibc/wiki/Testing/Testsuite Indeed not very handy. > Some tests requires a minimal amount of RAM that is not always provided by Qemu > target. > There is also libc-test project from musl project: > https://wiki.musl-libc.org/libc-test.html Interesting, although not sure how relevant is for glibc (would require to compare with old release). > Indeed running some tests from ltp-testsuite would be better than nothing. > On the other hand we can detect enough boot issue related to toolchain upgrade > to keep us busy. Thanks for all suggestions. I'm not sure when I get to it, but the max which time allows will probably be just to boot with qemu and run LTP syscalls. Kind regards, Petr > Best regards, > Romain > > Thomas
Hello Petr, Le 18/08/2021 à 23:53, Petr Vorel a écrit : > Hi Romain, > >> Hello, > >> Le 04/08/2021 à 21:30, Thomas Petazzoni a écrit : >>> Hello Petr, > >>> On Wed, 4 Aug 2021 19:22:59 +0200 >>> Petr Vorel <petr.vorel@gmail.com> wrote: > >>>> Also upgrade to on RISC-V (that could have been done in d5ce3b5db5) > >>>> Signed-off-by: Petr Vorel <petr.vorel@gmail.com> >>>> --- >>>> Hi, > >>>> I tested it with building pc_x86_64_efi_defconfig, adding nfs-utils, >>>> ltp-testsuite and few other tools. Not sure if this is enough for >>>> testing. > >>> I think that what Romain Naour typically does is run all qemu_* >>> defconfigs (of course tweaked to use glibc instead of uclibc), and push >>> them to Gitlab, so that they get build tested *and* boot tested under >>> Qemu. > >> Indeed, the runtime test is only a boot test. I've looked at the glibc testsuite >> but it's difficult to setup with a cross-compiled system running under Qemu in a >> CI (glitlab). > I hoped there would be some setup ready, but understand there is nothing. > >> For testing, upstream suggest to use a NFS to share the glibc sources between >> the host and the target (not very handy): >> https://sourceware.org/glibc/wiki/Testing/Testsuite > Indeed not very handy. > >> Some tests requires a minimal amount of RAM that is not always provided by Qemu >> target. > >> There is also libc-test project from musl project: >> https://wiki.musl-libc.org/libc-test.html > Interesting, although not sure how relevant is for glibc (would require to > compare with old release). > >> Indeed running some tests from ltp-testsuite would be better than nothing. >> On the other hand we can detect enough boot issue related to toolchain upgrade >> to keep us busy. > > Thanks for all suggestions. I'm not sure when I get to it, but the max which > time allows will probably be just to boot with qemu and run LTP syscalls. Please upgrade localedef package with the glibc version bump. I'll mark your patch "Changes requested". Best regards, Romain > > Kind regards, > Petr > >> Best regards, >> Romain > > >>> Thomas > >
Hi Romain, > Please upgrade localedef package with the glibc version bump. > I'll mark your patch "Changes requested". Sure, I'll send v2. Sorry for overlooking that comment suggesting it. Kind regards, Petr
diff --git a/package/glibc/2.32.9000-69-gbd394d131c10c9ec22c6424197b79410042eed99/glibc.hash b/package/glibc/2.32.9000-69-gbd394d131c10c9ec22c6424197b79410042eed99/glibc.hash deleted file mode 100644 index 558591522e..0000000000 --- a/package/glibc/2.32.9000-69-gbd394d131c10c9ec22c6424197b79410042eed99/glibc.hash +++ /dev/null @@ -1,7 +0,0 @@ -# Locally calculated (fetched from Github) -sha256 c3828224d69a42f5b6982cbec8ae01cf494fe975ce9023e36ff2c10a2fb93e42 glibc-2.32.9000-69-gbd394d131c10c9ec22c6424197b79410042eed99.tar.gz - -# Hashes for license files -sha256 8177f97513213526df2cf6184d8ff986c675afb514d4e68a404010521b880643 COPYING -sha256 dc626520dcd53a22f727af3ee42c770e56c97a64fe3adb063799d8ab032fe551 COPYING.LIB -sha256 b33d0bd9f685b46853548814893a6135e74430d12f6d94ab3eba42fc591f83bc LICENSES diff --git a/package/glibc/2.33-46-gedfd11197ecf3629bbb4b66c5814da09a61a7f9f/glibc.hash b/package/glibc/2.34-5-g9995d0588f4f9adc68419224d2b3698e2ca4f77e/glibc.hash similarity index 70% rename from package/glibc/2.33-46-gedfd11197ecf3629bbb4b66c5814da09a61a7f9f/glibc.hash rename to package/glibc/2.34-5-g9995d0588f4f9adc68419224d2b3698e2ca4f77e/glibc.hash index 4b8d8297ec..8b5aacb535 100644 --- a/package/glibc/2.33-46-gedfd11197ecf3629bbb4b66c5814da09a61a7f9f/glibc.hash +++ b/package/glibc/2.34-5-g9995d0588f4f9adc68419224d2b3698e2ca4f77e/glibc.hash @@ -1,5 +1,5 @@ # Locally calculated (fetched from Github) -sha256 3e2004b16ec6b79a19a88ed4736719ef9e419f9d8e7013faced654c0339e23dc glibc-2.33-46-gedfd11197ecf3629bbb4b66c5814da09a61a7f9f.tar.gz +sha256 77af51b251591d013ad4634863a42ac8ce62f4b2c7c29cba38a2ad5098d2727f glibc-2.34-5-g9995d0588f4f9adc68419224d2b3698e2ca4f77e.tar.gz # Hashes for license files sha256 8177f97513213526df2cf6184d8ff986c675afb514d4e68a404010521b880643 COPYING diff --git a/package/glibc/glibc.mk b/package/glibc/glibc.mk index 1c8c409878..320848bd01 100644 --- a/package/glibc/glibc.mk +++ b/package/glibc/glibc.mk @@ -11,13 +11,7 @@ else # Generate version string using: # git describe --match 'glibc-*' --abbrev=40 origin/release/MAJOR.MINOR/master | cut -d '-' -f 2- # When updating the version, please also update localedef -ifeq ($(BR2_RISCV_32),y) -# RISC-V 32-bit (RV32) requires glibc 2.33 or newer -# Until 2.33 is released, just use master -GLIBC_VERSION = 2.32.9000-69-gbd394d131c10c9ec22c6424197b79410042eed99 -else -GLIBC_VERSION = 2.33-46-gedfd11197ecf3629bbb4b66c5814da09a61a7f9f -endif +GLIBC_VERSION = 2.34-5-g9995d0588f4f9adc68419224d2b3698e2ca4f77e # Upstream doesn't officially provide an https download link. # There is one (https://sourceware.org/git/glibc.git) but it's not reliable, # sometimes the connection times out. So use an unofficial github mirror.
Also upgrade to on RISC-V (that could have been done in d5ce3b5db5) Signed-off-by: Petr Vorel <petr.vorel@gmail.com> --- Hi, I tested it with building pc_x86_64_efi_defconfig, adding nfs-utils, ltp-testsuite and few other tools. Not sure if this is enough for testing. Kind regards, Petr .../glibc.hash | 7 ------- .../glibc.hash | 2 +- package/glibc/glibc.mk | 8 +------- 3 files changed, 2 insertions(+), 15 deletions(-) delete mode 100644 package/glibc/2.32.9000-69-gbd394d131c10c9ec22c6424197b79410042eed99/glibc.hash rename package/glibc/{2.33-46-gedfd11197ecf3629bbb4b66c5814da09a61a7f9f => 2.34-5-g9995d0588f4f9adc68419224d2b3698e2ca4f77e}/glibc.hash (70%)