[Rpm-maint] [rpm-software-management/rpm] Packaging a fifo with rpm 4.18rc1 seems to hang (Issue #2195)
With rpm 4.18rc1 we've noticed that dnf seems to hang when installing a package
containing a fifo. The hang is from the new openat() call within fsmOpenat()
which dnf is calling from librpm.
Changing the condition in rpmPackageFilesInstall() from:
if (!rc && fd == -1 && !S_ISLNK(fp->sb.st_mode)) {
to
if (!rc && fd == -1 && !S_ISLNK(fp->sb.st_mode) && !S_ISFIFO(fp->sb.st_mode)) {
seems to avoid it.
In the interests of full disclosure, this was reproduced in Yocto Project and
was within our fakeroot emulation so it is possible there is some kind of bad
interaction happening there. I've not tried to reproduce outside our fakeroot
environment as yet as that isn't easy for us but I'm guessing you can test that
more easily and may want to fix this before 4.18 is released if it is a real
issue.
--
Reply to this email directly or view it on GitHub:
https://github.com/rpm-software-management/rpm/issues/2195
You are receiving this because you are subscribed to this thread.
Message ID: ___
Rpm-maint mailing list
Rpm-maint@lists.rpm.org
http://lists.rpm.org/mailman/listinfo/rpm-maint
Re: [Rpm-maint] [rpm-software-management/rpm] Fix a problem where scripts would fail to run with exit code 127 (#447)
Reworded the commit message. -- You are receiving this because you are subscribed to this thread. Reply to this email directly or view it on GitHub: https://github.com/rpm-software-management/rpm/pull/447#issuecomment-395362253___ Rpm-maint mailing list Rpm-maint@lists.rpm.org http://lists.rpm.org/mailman/listinfo/rpm-maint
Re: [Rpm-maint] [rpm-software-management/rpm] Fix a problem where scripts would fail to run with exit code 127 (#447)
Reopened #447. -- You are receiving this because you are subscribed to this thread. Reply to this email directly or view it on GitHub: https://github.com/rpm-software-management/rpm/pull/447#event-1668108516___ Rpm-maint mailing list Rpm-maint@lists.rpm.org http://lists.rpm.org/mailman/listinfo/rpm-maint
Re: [Rpm-maint] [rpm-software-management/rpm] Fix a problem where scripts would fail to run with exit code 127 (#447)
I realised there was a block further back in the code which sets xx =
setenv("PATH", path, 1); unconditionally so at the very least my patch
description is wrong. I think there is some questionable code here and would
welcome an opinion on what it should be doing.
In our local usage in Yocto Project, we actually have the setenv disabled as we
need our own PATH which is why we saw a much bigger problem and also see
compiler warnings. I'll reopen to trigger some further discussion but the patch
description is wrong as it stands. Happy to rewrite the description if that
helps and you're otherwise happy with the patch.
--
You are receiving this because you are subscribed to this thread.
Reply to this email directly or view it on GitHub:
https://github.com/rpm-software-management/rpm/pull/447#issuecomment-395359962___
Rpm-maint mailing list
Rpm-maint@lists.rpm.org
http://lists.rpm.org/mailman/listinfo/rpm-maint
Re: [Rpm-maint] [rpm-software-management/rpm] Fix a problem where scripts would fail to run with exit code 127 (#447)
Closed #447. -- You are receiving this because you are subscribed to this thread. Reply to this email directly or view it on GitHub: https://github.com/rpm-software-management/rpm/pull/447#event-1667062111___ Rpm-maint mailing list Rpm-maint@lists.rpm.org http://lists.rpm.org/mailman/listinfo/rpm-maint
[Rpm-maint] [rpm-software-management/rpm] Fix a problem where scripts would fail to run with exit code 127 (#447)
Commit 9c3e5de3240554c8ea1b29d52eeadee4840fefac, "Factor out and unify setting CLOEXEC" moved code which set the xx variable. There are now ways this function can fail to set xx at all leaving it at the mercy of the compiler. gcc 4.8.5 for example will leave this with a non-zero value causing scripts to fail. The original code here checked for return codes of functions which no longer set xx. As far as I can tell the best thing is to simple remove the seemingly erroneous test and make things deterministic! Signed-off-by: Richard Purdie You can view, comment on, or merge this pull request online at: https://github.com/rpm-software-management/rpm/pull/447 -- Commit Summary -- * Fix a problem where scripts would fail to run with exit code 127 -- File Changes -- M lib/rpmscript.c (4) -- Patch Links -- https://github.com/rpm-software-management/rpm/pull/447.patch https://github.com/rpm-software-management/rpm/pull/447.diff -- You are receiving this because you are subscribed to this thread. Reply to this email directly or view it on GitHub: https://github.com/rpm-software-management/rpm/pull/447 ___ Rpm-maint mailing list Rpm-maint@lists.rpm.org http://lists.rpm.org/mailman/listinfo/rpm-maint
Re: [Rpm-maint] [rpm-software-management/rpm] Do not hardcode "lib/rpm" as the installation path for default configuration and macros. (#298)
The Yocto Project doesn't mandate any particular userspace layout. You can configure it to match FHS but we expect software to match the configuration that is set. The issue here is that rpm doesn't honour our configuration settings. People can and do change the layouts for various reasons, I don't pass judgement on whether these are good or bad, our policy is to be flexible and do what users configure. If we can't change rpm in ways which let us pass configuration in, we'll just have to carry patches but obviously we'd prefer to avoid that if we can. -- You are receiving this because you are subscribed to this thread. Reply to this email directly or view it on GitHub: https://github.com/rpm-software-management/rpm/pull/298#issuecomment-319657461___ Rpm-maint mailing list Rpm-maint@lists.rpm.org http://lists.rpm.org/mailman/listinfo/rpm-maint
