Bug#624263: grub-pc: stuck in grub-mkdevicemap at configure time

[Jonathan Nieder]

Hi Yann,

Yann Dirson wrote:

> Got it again, when upgrading to 1.99-9, running under a 2.6.38.4
> kernel.  kern.log reads the following, maybe it will give more ideas
> about what's going on ?  Concurency issue in kernel space ?

Yeah, looks like it.  Next time you get a chance in this state, could
you try something like the following?

	echo d >/proc/sysrq-trigger; # list all locks held
	# or:
	echo w >/proc/sysrq-trigger; # list all blocked tasks

Also, is it reproducible (e.g., with "dpkg-reconfigure grub-pc")?
If so, trying a more recent kernel to compare results would be
interesting, too.

Trace left unsnipped for the kernel team's benefit.

Jonathan

> INFO: task grub-mkdevicema:24535 blocked for more than 120 seconds.
> "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
> grub-mkdevicema D 00000001034e66cc     0 24535  24534 0x00000000
>  ffff88008e95db28 0000000000000082 ffff880000000000 00000000000134c0
>  00000000000134c0 00000000000134c0 00000000000134c0 ffff880001ad5100
>  00000000000134c0 ffff88008e95dfd8 ffff88008e95dfd8 00000000000134c0
> Call Trace:
>  [<ffffffff8132e6ff>] __mutex_lock_common.clone.5+0x131/0x198
>  [<ffffffff811928a0>] ? exact_match+0x0/0xa
>  [<ffffffff8132e774>] __mutex_lock_slowpath+0xe/0x10
>  [<ffffffff8132e5b4>] mutex_lock+0x1e/0x38
>  [<ffffffff8112764c>] __blkdev_get+0x71/0x339
>  [<ffffffff81127ba3>] ? blkdev_open+0x0/0x6b
>  [<ffffffff81127ac8>] blkdev_get+0x1b4/0x28f
>  [<ffffffff81127ba3>] ? blkdev_open+0x0/0x6b
>  [<ffffffff81127c0a>] blkdev_open+0x67/0x6b
>  [<ffffffff810fddae>] __dentry_open+0x15e/0x278
>  [<ffffffff8132f14c>] ? _raw_spin_lock+0x9/0xb
>  [<ffffffff81115b82>] ? mntget+0x1b/0x21
>  [<ffffffff810feb2c>] nameidata_to_filp+0x5b/0x62
>  [<ffffffff8110a68d>] finish_open+0x9c/0x15a
>  [<ffffffff8110982a>] ? do_path_lookup+0xe8/0x113
>  [<ffffffff8110abb4>] do_filp_open+0x184/0x696
>  [<ffffffff810dd1a1>] ? handle_mm_fault+0x131/0x146
>  [<ffffffff811a5f39>] ? might_fault+0x9/0xb
>  [<ffffffff811a6025>] ? __strncpy_from_user+0x1a/0x49
>  [<ffffffff8111488a>] ? alloc_fd+0x6f/0x11a
>  [<ffffffff810feb8e>] do_sys_open+0x5b/0xed
>  [<ffffffff810fec3b>] sys_open+0x1b/0x1d
>  [<ffffffff81009992>] system_call_fastpath+0x16/0x1b