[Qemu-devel] qemu:handle_cpu_signal received signal outside vCPU context

[Qemu-devel] qemu:handle_cpu_signal received signal outside vCPU context

[Andreas Schwab]

I'm seeing this error while building gedit for riscv64 with linux-user
emulation:

$ LD_LIBRARY_PATH=gedit/.libs qemu-riscv64 gedit/.libs/gedit --introspect-dump=/tmp/tmp-introspectnj0xla07/functions.txt,/tmp/tmp-introspectnj0xla07/dump.xml
qemu:handle_cpu_signal received signal outside vCPU context @ pc=0x6003d7d5
qemu:handle_cpu_signal received signal outside vCPU context @ pc=0x60106a16

This is qemu as of today with the patches from
git://github.com/riscv/riscv-qemu / riscv-all on top.

Andreas.

-- 
Andreas Schwab, SUSE Labs, schwab@suse.de
GPG Key fingerprint = 0196 BAD8 1CE9 1970 F4BE  1748 E4D4 88E3 0EEA B9D7
"And now for something completely different."

Re: [Qemu-devel] qemu:handle_cpu_signal received signal outside vCPU context

[Michael Clark]

On Mon, Mar 19, 2018 at 9:17 AM, Andreas Schwab <schwab@suse.de> wrote:

> I'm seeing this error while building gedit for riscv64 with linux-user
> emulation:
>
> $ LD_LIBRARY_PATH=gedit/.libs qemu-riscv64 gedit/.libs/gedit
> --introspect-dump=/tmp/tmp-introspectnj0xla07/functions.txt,
> /tmp/tmp-introspectnj0xla07/dump.xml
> qemu:handle_cpu_signal received signal outside vCPU context @ pc=0x6003d7d5
> qemu:handle_cpu_signal received signal outside vCPU context @ pc=0x60106a16
>

Thanks. I can see this code in accel/tcg/user-exec.c

It would be nice if that log message included the signal number. I wonder
if we are getting a SIGSEGV. I also wonder what thread is actually
running...


> This is qemu as of today with the patches from
> git://github.com/riscv/riscv-qemu / riscv-all on top.


Okay. We don't have any linux-user specific patches in our patch series so
it likely the same in the master branch.

I wonder what is the best way for me to reproduce on my side... a tarball
with binaries that I can use to trigger the fault?

Re: [Qemu-devel] qemu:handle_cpu_signal received signal outside vCPU context

[Andreas Schwab]

On Mär 19 2018, Michael Clark <mjc@sifive.com> wrote:

> I wonder what is the best way for me to reproduce on my side... a tarball
> with binaries that I can use to trigger the fault?

The easiest way, if you have an OBS account, is to check out the package
at <https://build.opensuse.org/package/show/openSUSE:Factory:RISCV/gedit>
with osc and use osc build.

Andreas.

-- 
Andreas Schwab, SUSE Labs, schwab@suse.de
GPG Key fingerprint = 0196 BAD8 1CE9 1970 F4BE  1748 E4D4 88E3 0EEA B9D7
"And now for something completely different."

Re: [Qemu-devel] qemu:handle_cpu_signal received signal outside vCPU context

[Andreas Schwab]

On Mär 19 2018, Michael Clark <mjc@sifive.com> wrote:

> On Mon, Mar 19, 2018 at 9:17 AM, Andreas Schwab <schwab@suse.de> wrote:
>
>> I'm seeing this error while building gedit for riscv64 with linux-user
>> emulation:
>>
>> $ LD_LIBRARY_PATH=gedit/.libs qemu-riscv64 gedit/.libs/gedit
>> --introspect-dump=/tmp/tmp-introspectnj0xla07/functions.txt,
>> /tmp/tmp-introspectnj0xla07/dump.xml
>> qemu:handle_cpu_signal received signal outside vCPU context @ pc=0x6003d7d5
>> qemu:handle_cpu_signal received signal outside vCPU context @ pc=0x60106a16
>>
>
> Thanks. I can see this code in accel/tcg/user-exec.c
>
> It would be nice if that log message included the signal number. I wonder
> if we are getting a SIGSEGV. I also wonder what thread is actually
> running...

A native build doesn't see any issues, so this looks like a genuine qemu
bug.

> I wonder what is the best way for me to reproduce on my side... a tarball
> with binaries that I can use to trigger the fault?

There are some images under
<http://download.opensuse.org/repositories/openSUSE:/Factory:/RISCV/images/>
that can be used as a base.

Andreas.

-- 
Andreas Schwab, SUSE Labs, schwab@suse.de
GPG Key fingerprint = 0196 BAD8 1CE9 1970 F4BE  1748 E4D4 88E3 0EEA B9D7
"And now for something completely different."