Re: Linux 5.19-rc2

[Guenter Roeck <linux@roeck-us.net>]

On Sun, Jun 12, 2022 at 04:36:20PM -0700, Linus Torvalds wrote:
> It's Sunday afternoon, so it must be time for another release candidate.
> 
> As usual, rc2 is fairly small, and there isn't really a ton going on
> here. Fixes spread all over, and the diffstat is fairly flat apart
> from a few notable things that had bigger changes: some amdgpu fixes,
> the xen stub iommu driver, and some ceph and zonefs fixes. The rest is
> all pretty much one- or few-liners spread out mostly everywhere.
> 
> And yes, because I expected the rc2 week to be fairly quiet, I did a
> system update on my workstation, and as a result I spent a day or two
> then sorting out most of the fallout from the resulting compiler
> upgrade to gcc-12. Some of it ended up being a bit heavy-handed, and
> we'll be tweaking things further. And some of it ends up being an
> actual compiler misfeature, but it's being discussed too and is
> limited to just one file on the 32-bit i386 side (and does not seem to
> result in any actual bad code, just excessive stack use).
> 
> Anyway, nothing hugely scary going on, let's just keep testing and
> fixing things.
> 

Build results:
	total: 149 pass: 146 fail: 3
Failed builds:
	arm:allmodconfig
	mips:loongson2k_defconfig
	powerpc:allmodconfig
Qemu test results:
	total: 489 pass: 489 fail: 0

---
arm:allmodconfig

ERROR: modpost: missing MODULE_LICENSE() in drivers/watchdog/gxp-wdt.o

The fix is in -next, so hopefully Wim will send a pull request soon.

---
mips:loongson2k_defconfig

drivers/irqchip/irq-loongson-liointc.c: In function 'liointc_chained_handle_irq':
drivers/irqchip/irq-loongson-liointc.c:60:20: error: implicit declaration of function 'cpu_logical_map'

Caused by commit fa84f89395e03 ("irqchip/loongson-liointc: Fix build error for LoongArch").

---
powerpc:allmodconfig

Error: External symbol 'memset' referenced from prom_init.c
powerpc64-linux-ld:
	drivers/gpu/drm/amd/amdgpu/../display/dc/dml/display_mode_lib.o uses hard float,
	drivers/gpu/drm/amd/amdgpu/../display/dc/dcn31/dcn31_resource.o uses soft float
and similar errors. No change since last week.

---
Runtime warning:

OF: amba_device_add() failed (-19) for /amba/smc@10100000
------------[ cut here ]------------
WARNING: CPU: 0 PID: 1 at lib/refcount.c:28 of_platform_bus_create+0x33c/0x3dc
refcount_t: underflow; use-after-free.

Fixed in -next.

There are also various UAF and "spinlock bad magic" BUG/WARNING logs in arm
boot tests. I don't know if those are new or if I just see them now because
I enabled KFENCE. Here is an (incomplete) sample.

[    0.430290] printk: console [ramoops-1] printing thread started
[    0.436219] BUG: spinlock bad magic on CPU#0, kdevtmpfs/23
[    0.436250]  lock: noop_backing_dev_info+0x6c/0x3b0, .magic: 00000000, .owner: <none>/-1, .owner_cpu: 0
[    0.436808] CPU: 0 PID: 23 Comm: kdevtmpfs Not tainted 5.19.0-rc2 #1

---

[    7.918940] ------------[ cut here ]------------
[    7.919062] WARNING: CPU: 0 PID: 0 at kernel/smp.c:913 smp_call_function_many_cond+0x38c/0x3a4
[    7.919875] Modules linked in:
[    7.920155] CPU: 0 PID: 0 Comm: swapper/0 Not tainted 5.19.0-rc2 #1
[    7.920248] Hardware name: ARM-Versatile Express
[    7.920458]  unwind_backtrace from show_stack+0x10/0x14
[    7.920549]  show_stack from dump_stack_lvl+0x50/0x6c
[    7.920585]  dump_stack_lvl from __warn+0xcc/0x19c
[    7.920614]  __warn from warn_slowpath_fmt+0x5c/0xb4
[    7.920644]  warn_slowpath_fmt from smp_call_function_many_cond+0x38c/0x3a4
[    7.920674]  smp_call_function_many_cond from smp_call_function+0x3c/0x50
[    7.920702]  smp_call_function from set_memory_valid+0x74/0x94
[    7.920733]  set_memory_valid from kfence_guarded_free+0x280/0x4bc
[    7.920767]  kfence_guarded_free from kmem_cache_free+0x2f4/0x394

I'll try to bisect as I find the time.

Guenter