diff options
| author | Cole Helbling <cole.helbling@determinate.systems> | 2023-02-16 09:58:18 -0800 |
|---|---|---|
| committer | Cole Helbling <cole.e.helbling@outlook.com> | 2023-02-16 10:06:24 -0800 |
| commit | 4e2079b96d281212b695ca557755909799f163ad (patch) | |
| tree | 1dbd2bbc1f0fcfb060d8072ce58c5d6df9205f46 /pkgs/development/python-modules/termplotlib/gnuplot-subprocess.patch | |
| parent | ac1f5b72a9e95873d1de0233fddcb56f99884b37 (diff) | |
linux_{5_15,6_1}: revert patch to fix Equinix Metal bonded networking with `ice` driver
Some Equinix Metal instances, such as a3.large.x86, m3.large.x86
(specific hardware revisions), and n3.large.x86, use the `ice` kernel
driver for their network cards, in conjunction with bonded devices.
However, this commit caused a regression where these bonded devices
would deadlock. This was initially reported by Jaroslav Pulchart on
the netdev mailing list[1], and there were follow-up patches from Dave
Ertman[2][3] that attempted to fix this but were not up to snuff for
various reasons[4].
Specifically, v2 of the patch ([3]) appears to fix the issue on some
devices (tested with 8086:159B network cards), while it is still broken
on others (such as an 8086:1593 network card).
We revert the patch exposing the issue until upstream has a working
solution in order to make Equinix Metal instances work reliably again.
[1]: https://lore.kernel.org/netdev/CAK8fFZ6A_Gphw_3-QMGKEFQk=sfCw1Qmq0TVZK3rtAi7vb621A@mail.gmail.com/
[2]: https://patchwork.ozlabs.org/project/intel-wired-lan/patch/20230111183145.1497367-1-david.m.ertman@intel.com/
[3]: https://patchwork.ozlabs.org/project/intel-wired-lan/patch/20230215191757.1826508-1-david.m.ertman@intel.com/
[4]: https://lore.kernel.org/netdev/cb31a911-ba80-e2dc-231f-851757cfd0b8@intel.com/T/#m6e53f8c43093693c10268140126abe99e082dc1c
Diffstat (limited to 'pkgs/development/python-modules/termplotlib/gnuplot-subprocess.patch')
0 files changed, 0 insertions, 0 deletions
