10:49 a.m.
This follows the removal of the system call from Linux 5.5. Having the
header and function just confuses configure checks that assume that if
the function is present, it will do something useful (it never did on
aarch64, and it's been many years since it worked on Fedora kernels on
other architectures).
Replacements are uname, getentropy, and reading from /proc, but I really
do not expect much fallout from this (famous last words).
The kernel headers still provide <linux/sysctl.h>, but that will
eventually disappear as well.
Thanks,
Florian
4:33 a.m.
On 15. 04. 20 17:49, Florian Weimer wrote:
This follows the removal of the system call from Linux 5.5. Having
the
header and function just confuses configure checks that assume that if
the function is present, it will do something useful (it never did on
aarch64, and it's been many years since it worked on Fedora kernels on
other architectures).
Replacements are uname, getentropy, and reading from /proc, but I really
do not expect much fallout from this (famous last words).
The kernel headers still provide <linux/sysctl.h>, but that will
eventually disappear as well.
swift-lang fails to build:
/builddir/build/BUILD/swift-source/swift-corelibs-foundation/CoreFoundation/PlugIn.subproj/CFBundle_InfoPlist.c:21:10:
fatal error: 'sys/sysctl.h' file not found
#include <sys/sysctl.h>
^~~~~~~~~~~~~~
--
Miro Hrončok
--
Phone: +420777974800
IRC: mhroncok
7:41 a.m.
Already fixed it and sending it to Rawhide now.
-----Original Message-----
From: Miro Hrončok <mhroncok(a)redhat.com>
Sent: Saturday, April 18, 2020 4:34 AM
To: devel(a)lists.fedoraproject.org; swift-lang-maintainers(a)fedoraproject.org
Subject: Re: <sys/sysctl.h> will disappear from rawhide glibc soon
On 15. 04. 20 17:49, Florian Weimer wrote:
This follows the removal of the system call from Linux 5.5. Having
the header and function just confuses configure checks that assume
that if the function is present, it will do something useful (it never
did on aarch64, and it's been many years since it worked on Fedora
kernels on other architectures).
Replacements are uname, getentropy, and reading from /proc, but I
really do not expect much fallout from this (famous last words).
The kernel headers still provide <linux/sysctl.h>, but that will
eventually disappear as well.
swift-lang fails to build:
/builddir/build/BUILD/swift-source/swift-corelibs-foundation/CoreFoundation/PlugIn.subproj/CFBundle_InfoPlist.c:21:10:
fatal error: 'sys/sysctl.h' file not found #include <sys/sysctl.h>
^~~~~~~~~~~~~~
--
Miro Hrončok
--
Phone: +420777974800
IRC: mhroncok
11:26 a.m.
Yes: https://github.com/apple/swift-corelibs-foundation/pull/2771
-----Original Message-----
From: Florian Weimer <fweimer(a)redhat.com>
Sent: Saturday, April 18, 2020 8:27 AM
To: Ron Olson <tachoknight(a)gmail.com>
Cc: 'Miro Hrončok' <mhroncok(a)redhat.com>; devel(a)lists.fedoraproject.org;
swift-lang-maintainers(a)fedoraproject.org
Subject: Re: <sys/sysctl.h> will disappear from rawhide glibc soon
* Ron Olson:
Already fixed it and sending it to Rawhide now.
Have you submitted the fix upstream?
Thanks,
Florian
10:40 a.m.
On Wed, Apr 15, 2020 at 5:50 PM Florian Weimer <fweimer(a)redhat.com> wrote:
This follows the removal of the system call from Linux 5.5. Having the
header and function just confuses configure checks that assume that if
the function is present, it will do something useful (it never did on
aarch64, and it's been many years since it worked on Fedora kernels on
other architectures).
Replacements are uname, getentropy, and reading from /proc, but I really
do not expect much fallout from this (famous last words).
The kernel headers still provide <linux/sysctl.h>, but that will
eventually disappear as well.
java-1.8.0-openjdk also fails to build due to the missing header on
rawhide - this is blocking some work towards OpenJDK 11.
Fabio
> Thanks,
> Florian
> _______________________________________________
> devel mailing list -- devel(a)lists.fedoraproject.org
> To unsubscribe send an email to devel-leave(a)lists.fedoraproject.org
> Fedora Code of Conduct:
https://docs.fedoraproject.org/en-US/project/code-of-conduct/
> List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines
> List Archives:
https://lists.fedoraproject.org/archives/list/devel@lists.fedoraproject.org
10:59 a.m.
* Fabio Valentini:
On Wed, Apr 15, 2020 at 5:50 PM Florian Weimer
<fweimer(a)redhat.com> wrote:
>
> This follows the removal of the system call from Linux 5.5. Having the
> header and function just confuses configure checks that assume that if
> the function is present, it will do something useful (it never did on
> aarch64, and it's been many years since it worked on Fedora kernels on
> other architectures).
>
> Replacements are uname, getentropy, and reading from /proc, but I really
> do not expect much fallout from this (famous last words).
>
> The kernel headers still provide <linux/sysctl.h>, but that will
> eventually disappear as well.
java-1.8.0-openjdk also fails to build due to the missing header on
rawhide - this is blocking some work towards OpenJDK 11.
Do you have a build log? Do you need help with fixing this?
I thought that OpenJDK builds with -Werror, so it should have already
trippeed on the deprecation warning?
Thanks,
Florian
11:12 a.m.
On Tue, May 5, 2020 at 6:01 PM Florian Weimer <fweimer(a)redhat.com> wrote:
* Fabio Valentini:
> On Wed, Apr 15, 2020 at 5:50 PM Florian Weimer <fweimer(a)redhat.com> wrote:
>>
>> This follows the removal of the system call from Linux 5.5. Having the
>> header and function just confuses configure checks that assume that if
>> the function is present, it will do something useful (it never did on
>> aarch64, and it's been many years since it worked on Fedora kernels on
>> other architectures).
>>
>> Replacements are uname, getentropy, and reading from /proc, but I really
>> do not expect much fallout from this (famous last words).
>>
>> The kernel headers still provide <linux/sysctl.h>, but that will
>> eventually disappear as well.
>
> java-1.8.0-openjdk also fails to build due to the missing header on
> rawhide - this is blocking some work towards OpenJDK 11.
Do you have a build log? Do you need help with fixing this?
I thought that OpenJDK builds with -Werror, so it should have already
trippeed on the deprecation warning?
I don't know why this wasn't noticed sooner. Looking at koschei:
https://koschei.fedoraproject.org/package/java-1.8.0-openjdk?
It started to fail with glibc-2.31.9000-9.fc33, which was the first
build without sys/sysctl.h.
It does look like it builds with -Werror, though ...
These are the error messages from the latest build log:
/builddir/build/BUILD/java-1.8.0-openjdk-1.8.0.242.b08-3.fc33.x86_64/openjdk/jdk/src/solaris/native/java/net/PlainDatagramSocketImpl.c:44:10:
fatal error: sys/sysctl.h: No such file or directory
44 | #include <sys/sysctl.h>
| ^~~~~~~~~~~~~~
compilation terminated.
/builddir/build/BUILD/java-1.8.0-openjdk-1.8.0.242.b08-3.fc33.x86_64/openjdk/jdk/src/solaris/native/java/net/PlainSocketImpl.c:46:10:
fatal error: sys/sysctl.h: No such file or directory
46 | #include <sys/sysctl.h>
| ^~~~~~~~~~~~~~
compilation terminated.
I'm only indirectly impacted by this since I'm trying to fix ~200 java
packages that will be broken with OpenJDK 11, but for a working
environment to do that, we'd need to build a modified
java-1.8.0-openjdk ...
Fabio
> Thanks,
> Florian
>
1446
days inactive
1466
days old
7 comments
4 participants
participants (4)
-
Fabio Valentini -
Florian Weimer -
Miro Hrončok -
Ron Olson