From: nick clifton on gitlab.com
https://gitlab.com/cki-project/kernel-ark/-/merge_requests/2749#note_1626555205
BZ 2245296 should now be fixed by binutuls-2.41-9.fc40
___
kernel mailing list -- kernel@lists.fedoraproject.org
To unsubscribe send an
From: Viktor Malik on gitlab.com
https://gitlab.com/cki-project/kernel-ark/-/merge_requests/2749#note_1621698879
Applied upstream, patches are here:
https://lore.kernel.org/bpf/169832762501.4440.5459269955038255670.git-
patchwork-not...@kernel.org/
Feel free to pick them up as fit.
From: Carlos O'Donell (Red Hat) on gitlab.com
https://gitlab.com/cki-project/kernel-ark/-/merge_requests/2749#note_1619241995
@dzickusrh We are definitely triggering some odd issues with the flags we're
passing down at this point. You've managed to get clang's bpf backend to emit
From: Don Zickus on gitlab.com
https://gitlab.com/cki-project/kernel-ark/-/merge_requests/2749#note_1619009596
@codonell-rh - throwing the ball back into your court for above issue ^^
___
kernel mailing list -- kernel@lists.fedoraproject.org
To
From: Viktor Malik on gitlab.com
https://gitlab.com/cki-project/kernel-ark/-/merge_requests/2749#note_1618369263
So the `strip` problem is really a bug which @tohojo also noticed. He opened a
bugzilla for that:
https://bugzilla.redhat.com/show_bug.cgi?id=2245296
From: Viktor Malik on gitlab.com
https://gitlab.com/cki-project/kernel-ark/-/merge_requests/2749#note_1617635466
Patches sent, added you to cc. I'd still wait a couple of days, if possible,
to see if upstream doesn't propose a different approach, to avoid carrying
non-upstream patches in ARK. But
From: Viktor Malik on gitlab.com
https://gitlab.com/cki-project/kernel-ark/-/merge_requests/2749#note_1617583855
Not yet, I wanted to resolve all the `-fPIE` problems first b/c we'll
eventually need to build `samples/bpf` in RHEL.
Anyways, I think that we're on a good path (see my last comment
From: Viktor Malik on gitlab.com
https://gitlab.com/cki-project/kernel-ark/-/merge_requests/2749#note_1617578830
Argh, compiler/linker arguments in BPF Makefiles are sometimes a mess.
But, since the error occurs when linking `libbpf.a` for `samples/bpf/bpftool`,
let's take the same approach as
From: Don Zickus on gitlab.com
https://gitlab.com/cki-project/kernel-ark/-/merge_requests/2749#note_1617007832
@vmalikrh - were you able to push patches upstream? we were looking to bring
them in prematurely for ARK until upstream merges them. This helps move
binutils forward.
From: Don Zickus on gitlab.com
https://gitlab.com/cki-project/kernel-ark/-/merge_requests/2749#note_1612950237
@vmalikrh thanks for the fixes! I tried your suggestions and got this new
error:
```plaintext
+ /usr/bin/make -s 'HOSTCFLAGS=-O2 -fexceptions -g -grecord-gcc-switches
-pipe -Wall
From: Viktor Malik on gitlab.com
https://gitlab.com/cki-project/kernel-ark/-/merge_requests/2749#note_1612448872
@dzickusrh @hertonrk-rh I'd suggest applying one more patch (in addition to
the above quoting change) to `samples/bpf/Makefile`:
```diff
diff --git a/samples/bpf/Makefile
From: Don Zickus on gitlab.com
https://gitlab.com/cki-project/kernel-ark/-/merge_requests/2749#note_1609196658
I guess I spoke too soon, moving the TPROGS_xxx before make still causes rpm
to fail. hmm...
___
kernel mailing list --
From: Don Zickus on gitlab.com
https://gitlab.com/cki-project/kernel-ark/-/merge_requests/2749#note_1609190998
@hertonrk-rh @vmalikrh - moving TPROGS_xxx to before make resolved my missing
includes issue. But I see what @vmalikrh means for that one target.
The rpm still builds, so I think we
From: Viktor Malik on gitlab.com
https://gitlab.com/cki-project/kernel-ark/-/merge_requests/2749#note_1609066156
Argh, it won't work b/c [samples/bpf/Makefile:315](https://gitlab.com/cki-
project/kernel-ark/-/blob/os-build/samples/bpf/Makefile?ref_type=heads#L315)
overrides `TPROGS_CFLAGS` for
From: Viktor Malik on gitlab.com
https://gitlab.com/cki-project/kernel-ark/-/merge_requests/2749#note_1609047113
Agreed, I just wasn't sure if this is acceptable for the specfile. Makefile
always uses `+=` so it should work without modification (except for the above
Don's patch to enable setting
From: Herton R. Krzesinski on gitlab.com
https://gitlab.com/cki-project/kernel-ark/-/merge_requests/2749#note_1609004744
The way to do it in this case should be:
```TPROGS_CFLAGS="%{?build_hostcflags}" %{make} %{?_smp_mflags} ARCH=$Arch V=1
M=samples/bpf/ VMLINUX_H="${RPM_VMLINUX_H}" || true```
From: Viktor Malik on gitlab.com
https://gitlab.com/cki-project/kernel-ark/-/merge_requests/2749#note_1608645136
I don't think that setting `TPROGS_CFLAGS` here is correct.
The reason is that `samples/bpf/Makefile` appends a couple of options to
`TPROGS_CFLAGS`, particularly several include
From: Viktor Malik on gitlab.com
https://gitlab.com/cki-project/kernel-ark/-/merge_requests/2749#note_1608094306
@dzickusrh I can do that. If you don't mind, I'll add you to CC so that you
can see when it's merged.
___
kernel mailing list --
From: Don Zickus on gitlab.com
https://gitlab.com/cki-project/kernel-ark/-/merge_requests/2749#note_1607846588
@codonell-rh @vmalikrh - patching samples/bpf/Makefile (add quotes around
TPROGS_LDFLAGS) allowed me to use TPROGS_LDFLAGS correctly.
@vmalikrh - who to send this patch to upstream? I
From: Don Zickus on gitlab.com
https://gitlab.com/cki-project/kernel-ark/-/merge_requests/2749#note_1607668819
@codonell-rh Sure. Just struggling to figure out how to workaround that.
___
kernel mailing list -- kernel@lists.fedoraproject.org
To
From: Carlos O'Donell (Red Hat) on gitlab.com
https://gitlab.com/cki-project/kernel-ark/-/merge_requests/2749#note_1607368686
This indicates incorrect shell quoting for the intended usage with options
being expanded directly into Make's command line instead of the eventual
expansion for the
From: Don Zickus on gitlab.com
https://gitlab.com/cki-project/kernel-ark/-/merge_requests/2749#note_1605950727
For some reason adding the TPRROGS_LDFLAGS creates an ld usage failure.
+ /usr/bin/make -s 'HOSTCFLAGS=-O2 -fexceptions -g -grecord-gcc-switches
-pipe -Wall -Werror=format-security
From: Carlos O'Donell (Red Hat) on gitlab.com
https://gitlab.com/cki-project/kernel-ark/-/merge_requests/2749#note_1602841029
Any reason not to pass TPROGS_LDFLAGS for consistency for test programs?
Otherwise looks good.
___
kernel mailing list --
From: Carlos O'Donell (Red Hat) on gitlab.com
https://gitlab.com/cki-project/kernel-ark/-/merge_requests/2749#note_1602841038
OK. LGTM. Set USERCFLAGS and USERLDFLAGS.
___
kernel mailing list -- kernel@lists.fedoraproject.org
To unsubscribe send an
From: Don Zickus on gitlab.com
https://gitlab.com/cki-project/kernel-ark/-/merge_requests/2749#note_1602503582
@vmalikrh - FYI as the approval rules won't automatically add you.
___
kernel mailing list -- kernel@lists.fedoraproject.org
To unsubscribe
From: Don Zickus
Fix binutils breakage
The binutils package in Rawhide has stricter checks about PIE/PIC code
and will start reporting errors if it detects mixing and matching no-PIE
with PIE binaries (especially around bpf binaries).
Example errors look like:
/usr/bin/ld: /tmp/ccL7dkfR.o:
26 matches
Mail list logo