Re: Linux 5.15-rc1

From: Guenter Roeck
Date: Sun Sep 12 2021 - 23:57:56 EST


On Sun, Sep 12, 2021 at 04:58:27PM -0700, Linus Torvalds wrote:
[ ... ]
>
> And in order to get those fixes going, please go out and test this.
>

Build results:
total: 153 pass: 143 fail: 10
Failed builds:
alpha:allmodconfig (15)
arm:allmodconfig (1)
m68k:allmodconfig (47)
mips:allmodconfig (2)
parisc:allmodconfig (8)
riscv32:allmodconfig (1)
riscv:allmodconfig (1)
s390:allmodconfig (6)
sparc64:allmodconfig (1)
xtensa:allmodconfig (53)
Qemu test results:
total: 479 pass: 478 fail: 1
Failed tests:
sparc64:sun4u:nodebug:smp:virtio-pci:net,i82559er:hd

The allmodconfig build failures are mostly if not all the result of
enabling -Werror on test builds. The individual errors are too many
to list; the number of errors is listed in () after the build above.
Anyone interested can find build logs at https://kerneltests.org/builders.
I am not entirely sure what to do about that - if history teaches a
lesson, new build failures may pile up faster than existing ones get
fixed, but then I am aware that several fixes are queued already. We'll
see how it goes. I'll monitor the situation until maybe -rc4 or -rc5 and
then decide if I disable -Werror in my test builds to catch any new "real"
problems.

The qemu runtime failure bisects to commit 694a1116b405 ("virtio: Bind
virtio device to device-tree node"), and reverting that commit fixes the
problem. With that patch applied, the virtio block device does not
instantiate on sparc64. This results in a crash since that is where the
test is trying to boot from.

Good news is that I don't see any new runtime warnings.

Guenter