Message-ID: <ebb0791c-60a5-4fe3-bc85-4921ad026605@heusipp.de>
Date: Thu, 30 May 2024 08:55:24 +0200
From: Jörn Heusipp <osmanx@...sipp.de>
To: linux-kernel@...r.kernel.org
Cc: regressions@...ts.linux.dev, x86@...nel.org,
 Thomas Gleixner <tglx@...utronix.de>, Ingo Molnar <mingo@...hat.com>,
 Dave Hansen <dave.hansen@...ux.intel.com>, Borislav Petkov <bp@...en8.de>,
 Andy Lutomirski <luto@...nel.org>, Peter Zijlstra <peterz@...radead.org>,
 Adam Dunlap <acdunlap@...gle.com>, 1071378@...s.debian.org
Subject: [REGRESSION] commit fbf6449f84bf5e4ad09f2c09ee70ed7d629b5ff6 (Linux
 6.7+) crashes during boot


Hello x86 maintainers!


commit fbf6449f84bf5e4ad09f2c09ee70ed7d629b5ff6 ("x86/sev-es: Set
x86_virt_bits to the correct value straight away, instead of a two-phase
approach") crashes during boot for me on this 32bit x86 system.

Updating a Debian testing system resulted in a hang during boot before
printing anything, with any 6.7 or later kernel. With 'earlyprintk=vga',
I managed to capture the crash on video and stitched it together as an
image [1].
Trimmed transcription (might contain typos) of the crash from Debian
kernel 6.7.12-1:
===
BUG: kernel NULL pointer dereference, address: 00000010
#PF: supervisor write access in kernel mode
#PF: error_code(0x0002) - not-present page
Oops: 0002 [#1] PREEMPT SMP NOPTI
[...]
EIP: __ring_buffer_alloc+0x32/0x194
[...]
show_regs
__die
page_fault_oops
kernelmode_fixup_or_oops.constprop
__bad_area_nosemaphore.constprop
bad_area_nosemaphore
do_user_addr_fault
prb_read_valid
exc_page_fault
pvclock_clocksource_read_nowd
handle_exception
pvclock_clocksource_read_nowd
__ring_buffer_alloc
pvclock_clocksource_read_nowd
__ring_buffer_alloc
early_trace_init
start_kernel
i386_start_kernel
startup_32_smp
[...]
===
I could transcribe all of it or capture it again from latest git and
decode the symbols, if truely really needed, but I figured the type of
crash and the trace itself could maybe be sufficient. It looks identical
to me for all later crashing kernel versions.

I bisected this down to commit fbf6449f84bf5e4ad09f2c09ee70ed7d629b5ff6.

The kernel config [2] I used is 'make olddefconfig' based on Debian's
config-6.8.11-686-pae [3].

I also tested 6.9.2 and 6.10-rc1, both also still crash in the same way.

cpuinfo:
===
manx@...sar:~$ cat /proc/cpuinfo
processor       : 0
vendor_id       : AuthenticAMD
cpu family      : 6
model           : 8
model name      : AMD Duron(tm)
stepping        : 1
cpu MHz         : 1798.331
cache size      : 64 KB
physical id     : 0
siblings        : 1
core id         : 0
cpu cores       : 1
apicid          : 0
initial apicid  : 0
fdiv_bug        : no
f00f_bug        : no
coma_bug        : no
fpu             : yes
fpu_exception   : yes
cpuid level     : 1
wp              : yes
flags           : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge
mca cmov pat pse36 mmx fxsr sse syscall mmxext 3dnowext 3dnow cpuid
3dnowprefetch vmmcall
bugs            : fxsave_leak sysret_ss_attrs spectre_v1 spectre_v2
spec_store_bypass
bogomips        : 3596.66
clflush size    : 32
cache_alignment : 32
address sizes   : 34 bits physical, 32 bits virtual
power management: ts
===

dmesg from a successful boot (Debian kernel 6.6.15-2) is here [4].

This particular system has been running all Debian testing kernels since
at least the 2.6.32 days and is currently running 6.6.15-2 completely
fine, thus this is an obvious regression.

The original Debian bug is #1071378 [5].


#regzbot introduced: fbf6449f84bf5e4ad09f2c09ee70ed7d629b5ff6

[1] https://manx.datengang.de/temp/linux-6.7-crash/6.7.12-1-crash.png
[2] https://manx.datengang.de/temp/linux-6.7-crash/config
[3] https://manx.datengang.de/temp/linux-6.7-crash/config-6.8.11-686-pae
[4] https://manx.datengang.de/temp/linux-6.7-crash/dmesg-6.6.15-2.txt
[5] https://bugs.debian.org/1071378


Best regards,
Jörn

Powered by blists - more mailing lists