RE: [OE-core][PATCH] openssl: upgrade 3.3.1 -> 3.4.0

["Marko, Peter" <Peter.Marko@siemens.com>]

Richard,

could this be queued in master-next again?
I have submitted fixes for python3 to support openssl 3.4.x.
https://lists.openembedded.org/g/openembedded-core/message/208030

Thanks,
  Peter

> -----Original Message-----
> From: Marko, Peter (FT D EU SK BFS1)
> Sent: Thursday, October 24, 2024 20:56
> To: Richard Purdie <richard.purdie@linuxfoundation.org>; openembedded-
> core@lists.openembedded.org; Steve Sakoman <steve@sakoman.com>
> Subject: RE: [OE-core][PATCH] openssl: upgrade 3.3.1 -> 3.4.0
> 
> > -----Original Message-----
> > From: Richard Purdie <richard.purdie@linuxfoundation.org>
> > Sent: Thursday, October 24, 2024 12:14
> > To: Marko, Peter (FT D EU SK BFS1) <Peter.Marko@siemens.com>;
> > openembedded-core@lists.openembedded.org; Steve Sakoman
> > <steve@sakoman.com>
> > Subject: Re: [OE-core][PATCH] openssl: upgrade 3.3.1 -> 3.4.0
> >
> > On Thu, 2024-10-24 at 09:19 +0100, Richard Purdie via
> > lists.openembedded.org wrote:
> > > On Wed, 2024-10-23 at 20:48 +0200, Peter Marko via
> > > lists.openembedded.org wrote:
> > > > From: Peter Marko <peter.marko@siemens.com>
> > > >
> > > > Release information:
> > > > https://github.com/openssl/openssl/blob/openssl-3.4/NEWS.md#major-
> > changes-between-openssl-33-and-openssl-340-22-oct-2024
> > > >
> > > > Handles CVE-2024-9143
> > > >
> > > > Refreshed patches.
> > > >
> > > > Signed-off-by: Peter Marko <peter.marko@siemens.com>
> > > > ---
> > > >  ...ke-history-reporting-when-test-fails.patch | 53 ++++++++-------
> > > > ----
> > > >  ...1-Configure-do-not-tweak-mips-cflags.patch |  4 +-
> > > >  ...sysroot-and-debug-prefix-map-from-co.patch |  4 +-
> > > >  .../{openssl_3.3.1.bb => openssl_3.4.0.bb}    |  2 +-
> > > >  4 files changed, 28 insertions(+), 35 deletions(-)
> > > >  rename meta/recipes-connectivity/openssl/{openssl_3.3.1.bb =>
> > > > openssl_3.4.0.bb} (99%)
> > >
> > > Unfortunately this causes a regression in python3's ptest:
> > >
> > > https://valkyrie.yoctoproject.org/#/builders/73/builds/270
> > >
> > > I did revert this patch and tested a build and that passes so it is
> > > definitely from here.
> > >
> > > I'm trying this in conjunction with the python 3.13 upgrade which
> > > might
> > > let us fix this for master but if so, the problem for styhead and
> > > earlier will probably remain.
> > >
> > > There was a related python bug:
> > >
> > > https://github.com/python/cpython/issues/87743
> > >
> > > Copying Steve so he knows that gremlins lurk here...
> >
> > The 3.13 upgrade doesn't help so we're going to need to resolve the
> > ptest issue somehow to merge this...
> >
> > Cheers,
> >
> > Richard
> 
> Richard,
> 
> I was able to bisect openssl to
> https://github.com/openssl/openssl/commit/933f57dfe21657f7aba8f13e0cdb3b0
> 2dd64fcc3
> When I revert this commit, then the python3-ptest succeeds also with openssl
> 3.4.0.
> I have created https://github.com/python/cpython/issues/125936 to inform
> cpython maintainers.
> I don't have enough knowledge to fix it myself but I think that the openssl commit
> breaking things should be great lead for someone else.
> 
> During the investigation, I have also found 2 following problems with python:
> 
> 1) when running ptest in core-image-minimal, there are two crashes in console
> (both with openssl 3.3.1 and 3.4.0):
> [   91.482461] python3[8605]: segfault at 10 ip 00007f5b0760f561 sp
> 00007ffeda12c440 error 4 in
> libpython3.12.so.1.0[1ae561,7f5b07566000+23c000] likely on CPU 0 (core 0,
> socket 0)
> [   91.485213] Code: 0f 1f 84 00 00 00 00 00 90 41 54 55 53 48 89 fb 66 48 8d 3d
> c9 d9 35 00 66 66 48 e8 79 90 f5 ff 4c 8b 25 ca d7 35 00 48 8b 00 <48> 8b 68 10
> 49 3b 6c 24 30 74 09 f6 85 b0 06 00 00 20 75 4b 48 8d
> [  118.871057] python3[12214]: segfault at 10 ip 00007f7ad5dd2561 sp
> 00007f7ad50f7dd0 error 4 in
> libpython3.12.so.1.0[1ae561,7f7ad5d29000+23c000] likely on CPU 2 (core 2,
> socket 0)
> [  118.873991] Code: 0f 1f 84 00 00 00 00 00 90 41 54 55 53 48 89 fb 66 48 8d 3d
> c9 d9 35 00 66 66 48 e8 79 90 f5 ff 4c 8b 25 ca d7 35 00 48 8b 00 <48> 8b 68 10
> 49 3b 6c 24 30 74 09 f6 85 b0 06 00 00 20 75 4b 48 8d
> 
> 2) python ptest parsing is wrong, it detected only one from 8 test failures (see my
> cpython issue and ptest logs).
> 
> Could someone create Bugzilla entries for these?
> 
> Regards,
>   Peter