Failure booting hurd with qemu-system-i386 on ARM

[PeteVine]

Trying to boot debian-hurd-20150320.img ends with:

qemu-system-i386: qemu-coroutine-lock.c:91: qemu_co_queue_restart_all: Assertion `qemu_in_coroutine()' failed.

Program received signal SIGABRT, Aborted.
__libc_do_syscall ()
    at ../ports/sysdeps/unix/sysv/linux/arm/libc-do-syscall.S:44
44 ../ports/sysdeps/unix/sysv/linux/arm/libc-do-syscall.S: No such file or directory.
(gdb) bt
#0 __libc_do_syscall ()
    at ../ports/sysdeps/unix/sysv/linux/arm/libc-do-syscall.S:44
#1 0xb6ef8f0e in __GI_raise (sig=sig@entry=6)
    at ../nptl/sysdeps/unix/sysv/linux/raise.c:56
#2 0xb6efb766 in __GI_abort () at abort.c:89
#3 0xb6ef4150 in __assert_fail_base (
    fmt=0x1 <error: Cannot access memory at address 0x1>,
    assertion=0x7f89a234 "qemu_in_coroutine()", assertion@entry=0x0,
    file=0x7f89da58 "qemu-coroutine-lock.c", file@entry=0xb5660000 "\001",
    line=91, line@entry=3069931692,
    function=function@entry=0x7f89ab78 "qemu_co_queue_restart_all")
    at assert.c:92
#4 0xb6ef41e6 in __GI___assert_fail (assertion=0x0, file=0xb5660000 "\001",
    line=3069931692, function=0x7f89ab78 "qemu_co_queue_restart_all")
    at assert.c:101
#5 0x7f59a6b4 in ?? ()

I was using the same setup as in Bug 893208 (i.e git checkout from 2015-09-15, armv7 Odroid C1)

[Laszlo Ersek (Red Hat) (lersek)]

On 09/21/15 21:04, PeteVine wrote:
> Public bug reported:
>
> Trying to boot debian-hurd-20150320.img ends with:
>
> qemu-system-i386: qemu-coroutine-lock.c:91: qemu_co_queue_restart_all:
> Assertion `qemu_in_coroutine()' failed.
>
> Program received signal SIGABRT, Aborted.
> __libc_do_syscall ()
> at ../ports/sysdeps/unix/sysv/linux/arm/libc-do-syscall.S:44
> 44 ../ports/sysdeps/unix/sysv/linux/arm/libc-do-syscall.S: No such file or directory.
> (gdb) bt
> #0 __libc_do_syscall ()
> at ../ports/sysdeps/unix/sysv/linux/arm/libc-do-syscall.S:44
> #1 0xb6ef8f0e in __GI_raise (sig=sig@entry=6)
> at ../nptl/sysdeps/unix/sysv/linux/raise.c:56
> #2 0xb6efb766 in __GI_abort () at abort.c:89
> #3 0xb6ef4150 in __assert_fail_base (
> fmt=0x1 <error: Cannot access memory at address 0x1>,
> assertion=0x7f89a234 "qemu_in_coroutine()", assertion@entry=0x0,
> file=0x7f89da58 "qemu-coroutine-lock.c", file@entry=0xb5660000 "\001",
> line=91, line@entry=3069931692,
> function=function@entry=0x7f89ab78 "qemu_co_queue_restart_all")
> at assert.c:92
> #4 0xb6ef41e6 in __GI___assert_fail (assertion=0x0, file=0xb5660000 "\001",
> line=3069931692, function=0x7f89ab78 "qemu_co_queue_restart_all")
> at assert.c:101
> #5 0x7f59a6b4 in ?? ()
>
> I was using the same setup as in Bug 893208 (i.e git checkout from
> 2015-09-15)
>
> ** Affects: qemu
> Importance: Undecided
> Status: New
>

This backtrace is next to useless I believe, but it's not your fault. I
think "scripts/qemu-gdb.py" might be helpful (it has a little bit of
documentation too). See also commit 9eddd6a4.

Thanks
Laszlo

[PeteVine (davine-k)]

All right, I'll rebuild and try qemu-gdb.py if still necessary.
Do any fancy CFLAGS make a difference in performance or should they be avoided altogether on ARM?

[PeteVine (davine-k)]

I can't get any more useful info - either the script is expecting some outdated version of python or there's simply nothing else to see cause qemu failed to start.

For example:

(gdb) source qemu-gdb.py
(gdb) run
Starting program: /usr/bin/qemu-system-i386 -m 512 -hda /media/odroid/debian-hurd-20150320.img

and so on

(gdb) qemu mtree
Python Exception <class 'gdb.error'> No symbol "address_space_memory" in current context.:
Error occurred in Python command: No symbol "address_space_memory" in current context.

As for using:

qemu coroutine, I'm not sure where the pointer value should come from but feeding it a bogus one ends exactly as above.

Any suggestions?

[Laszlo Ersek (Red Hat) (lersek)]

On 09/23/15 17:13, PeteVine wrote:
> I can't get any more useful info - either the script is expecting some
> outdated version of python or there's simply nothing else to see cause
> qemu failed to start.
>
> For example:
>
> (gdb) source qemu-gdb.py
> (gdb) run
> Starting program: /usr/bin/qemu-system-i386 -m 512 -hda /media/odroid/debian-hurd-20150320.img
>
> and so on
>
> (gdb) qemu mtree
> Python Exception <class 'gdb.error'> No symbol "address_space_memory" in current context.:
> Error occurred in Python command: No symbol "address_space_memory" in current context.
>
> As for using:
>
> qemu coroutine, I'm not sure where the pointer value should come from
> but feeding it a bogus one ends exactly as above.
>
> Any suggestions?
>

Sorry, no idea.

[PeteVine (davine-k)]

I've just tried again with the latest commits - hurd boots, hooray!

[PeteVine (davine-k)]

Even though not related to the original issue (was also happening on i386 a few days ago), after getting to the login prompt inside hurd the keyboard doesn't work and the only clue from the kernel at boot time might be this:

'Unexpected ACK from keyboard'

or this:

'/bin/console: could not receive return value from the daemon process: Connection timed out'

I haven't got any more info so I'm not going to open a new bug myself. Thx.

[PeteVine (davine-k)]

Lastly, the machine 'power down' button doesn't work and a new message appeared inside hurd:

'kdb: queue full'

[PeteVine (davine-k)]

In case anyone's interested I've just discovered booting in recovery mode (root already logged in) doesn't exhibit the problem with non-working keyboard.

[Thomas Huth (th-huth)]

Looking through old bug tickets... can you still reproduce this issue with the latest version of QEMU? Or could we close this ticket nowadays?

[Launchpad Janitor (janitor)]

[Expired for QEMU because there has been no activity for 60 days.]