[Qemu-devel] [ARM][regression][bisect] ARM target broken: test v5 image does not start kernel

[Claudio Fontana]

Hello all,

I just finished bisecting a regression I am experiencing on ARM 32bit target,
while testing qemu-system-arm built for the aarch64 host running the ARMv5 
integrator image from our test page at

http://wiki.qemu.org/Testing

The breakage assumes the form of failure to start the kernel after

Uncompressing Linux.... done, booting the kernel.

With info registers showing registers stuck, with PC=c002f6bc.
Worked fine in July.

Can you help me get this to work again?

After a painful bisection, I got a first bad commit, which when reverted on the 
latest QEMU fixes the issue for me. Maybe something needs to be adapted to the 
return value (back-)change?

Bisection logs follow.

Thank you all,

Claudio

9b8c69243585a32d14b9bb9fcd52c37b0b5a1b71 is the first bad commit
commit 9b8c69243585a32d14b9bb9fcd52c37b0b5a1b71
Author: Jan Kiszka <address@hidden>
Date:   Tue Jul 16 14:45:16 2013 +0200

    memory: Return -1 again on reads from unsigned regions

    This restore the behavior prior to b018ddf633 which accidentally changed
    the return code to 0. Specifically guests probing for register existence
    were affected by this.

    Signed-off-by: Jan Kiszka <address@hidden>
    Signed-off-by: Paolo Bonzini <address@hidden>

:100644 100644 9938b6ba458e6e3caeb18d64b9fac2c64bc04c34 
34a088e90f4cb745e6fd462c6fc9b14bc6b25f41 M      memory.c

git bisect start
# bad: [aaa6a40194e9f204cb853f64ef3c1e170bb014e8] Merge remote-tracking branch 
'afaerber/tags/qom-cpu-for-anthony' into staging
git bisect bad aaa6a40194e9f204cb853f64ef3c1e170bb014e8
# good: [1acd5a373905ddb28957842256a038956941f332] Merge remote-tracking branch 
'agraf/ppc-for-upstream' into staging
git bisect good 1acd5a373905ddb28957842256a038956941f332
# bad: [f9e741903982e55c0dc138ab2e61059a4f3c9a76] cs4231: QOM cast cleanup
git bisect bad f9e741903982e55c0dc138ab2e61059a4f3c9a76
# bad: [f9e741903982e55c0dc138ab2e61059a4f3c9a76] cs4231: QOM cast cleanup
git bisect bad f9e741903982e55c0dc138ab2e61059a4f3c9a76
# bad: [c9fea5d701f8fd33f0843728ec264d95cee3ed37] Merge remote-tracking branch 
'bonzini/iommu-for-anthony' into staging
git bisect bad c9fea5d701f8fd33f0843728ec264d95cee3ed37
# good: [91b1df8cf9e1ecaa8679c9ea8713d1e25c28e6c4] cpu: Move reset logging to 
CPUState
git bisect good 91b1df8cf9e1ecaa8679c9ea8713d1e25c28e6c4
# good: [91b1df8cf9e1ecaa8679c9ea8713d1e25c28e6c4] cpu: Move reset logging to 
CPUState
git bisect good 91b1df8cf9e1ecaa8679c9ea8713d1e25c28e6c4
# good: [a34001fab5da2d0df605a8b83880c917c8aa0606] Merge remote-tracking branch 
'rth/axp-next' into staging
git bisect good a34001fab5da2d0df605a8b83880c917c8aa0606
# good: [a34001fab5da2d0df605a8b83880c917c8aa0606] Merge remote-tracking branch 
'rth/axp-next' into staging
git bisect good a34001fab5da2d0df605a8b83880c917c8aa0606
# good: [fdd26fca3ce66863e547560fbde1a444fc5d71b7] libqtest: Plug fd and memory 
leaks in qtest_quit()
git bisect good fdd26fca3ce66863e547560fbde1a444fc5d71b7
# good: [293706dd682f578b457d052988cf3c20b4eab82d] Merge remote-tracking branch 
'rth/axp-next' into staging
git bisect good 293706dd682f578b457d052988cf3c20b4eab82d
# good: [293706dd682f578b457d052988cf3c20b4eab82d] Merge remote-tracking branch 
'rth/axp-next' into staging
git bisect good 293706dd682f578b457d052988cf3c20b4eab82d
# good: [a23fdf355969d331f60593fa5b857d43aec25aac] block/raw: add .bdrv_get_info
git bisect good a23fdf355969d331f60593fa5b857d43aec25aac
# good: [be022d61f4938bb051e8af8e6cb470ec1602c488] doc: monitor multiplexing 
rewording
git bisect good be022d61f4938bb051e8af8e6cb470ec1602c488
# bad: [9b8c69243585a32d14b9bb9fcd52c37b0b5a1b71] memory: Return -1 again on 
reads from unsigned regions
git bisect bad 9b8c69243585a32d14b9bb9fcd52c37b0b5a1b71
# good: [b4afea11aafe85975e74dd562bb94f7ce3de1ef1] memory: actually set the 
owner
git bisect good b4afea11aafe85975e74dd562bb94f7ce3de1ef1


-- 
Claudio Fontana
Server OS Architect
Huawei Technologies Duesseldorf GmbH