[PATCH] kexec: add cond_resched into kimage_alloc_crash_control_pages

[Eric W. Biederman]

zhongjiang <zhongjiang at huawei.com> writes:

> From: zhong jiang <zhongjiang at huawei.com>
>
> A soft lookup will occur when I run trinity in syscall kexec_load.
> the corresponding stack information is as follows.

Overall that looks reasonable.  Why only every 256 page and not call
cond_resched unconditionally?

The function cond_resched won't reschedule unless the process has spent
it's cpu quota anyway.

Eric

> [  237.235937] BUG: soft lockup - CPU#6 stuck for 22s! [trinity-c6:13859]
> [  237.242699] Kernel panic - not syncing: softlockup: hung tasks
> [  237.248573] CPU: 6 PID: 13859 Comm: trinity-c6 Tainted: G           O L ----V-------   3.10.0-327.28.3.35.zhongjiang.x86_64 #1
> [  237.259984] Hardware name: Huawei Technologies Co., Ltd. Tecal BH622 V2/BC01SRSA0, BIOS RMIBV386 06/30/2014
> [  237.269752]  ffffffff8187626b 0000000018cfde31 ffff88184c803e18 ffffffff81638f16
> [  237.277471]  ffff88184c803e98 ffffffff8163278f 0000000000000008 ffff88184c803ea8
> [  237.285190]  ffff88184c803e48 0000000018cfde31 ffff88184c803e67 0000000000000000
> [  237.292909] Call Trace:
> [  237.295404]  <IRQ>  [<ffffffff81638f16>] dump_stack+0x19/0x1b
> [  237.301352]  [<ffffffff8163278f>] panic+0xd8/0x214
> [  237.306196]  [<ffffffff8111d6fc>] watchdog_timer_fn+0x1cc/0x1e0
> [  237.312157]  [<ffffffff8111d530>] ? watchdog_enable+0xc0/0xc0
> [  237.317955]  [<ffffffff810aa182>] __hrtimer_run_queues+0xd2/0x260
> [  237.324087]  [<ffffffff810aa720>] hrtimer_interrupt+0xb0/0x1e0
> [  237.329963]  [<ffffffff8164ae5c>] ? call_softirq+0x1c/0x30
> [  237.335500]  [<ffffffff81049a77>] local_apic_timer_interrupt+0x37/0x60
> [  237.342228]  [<ffffffff8164bacf>] smp_apic_timer_interrupt+0x3f/0x60
> [  237.348771]  [<ffffffff8164a11d>] apic_timer_interrupt+0x6d/0x80
> [  237.354967]  <EOI>  [<ffffffff810f3a00>] ? kimage_alloc_control_pages+0x80/0x270
> [  237.362875]  [<ffffffff811c3ebe>] ? kmem_cache_alloc_trace+0x1ce/0x1f0
> [  237.369592]  [<ffffffff810f362f>] ? do_kimage_alloc_init+0x1f/0x90
> [  237.375992]  [<ffffffff810f3d1a>] kimage_alloc_init+0x12a/0x180
> [  237.382103]  [<ffffffff810f3f9a>] SyS_kexec_load+0x20a/0x260
> [  237.387957]  [<ffffffff816494c9>] system_call_fastpath+0x16/0x1b
>
> the first time allocate control pages may take too much time because
> crash_res.end can be set to a higher value. we need to add cond_resched
> to avoid the issue.
>
> The patch have been tested and above issue is not appear.
>
> Signed-off-by: zhong jiang <zhongjiang at huawei.com>
> ---
>  kernel/kexec_core.c | 4 ++++
>  1 file changed, 4 insertions(+)
>
> diff --git a/kernel/kexec_core.c b/kernel/kexec_core.c
> index 5616755..2b43cc5 100644
> --- a/kernel/kexec_core.c
> +++ b/kernel/kexec_core.c
> @@ -433,6 +433,7 @@ static struct page *kimage_alloc_crash_control_pages(struct kimage *image,
>  	 */
>  	unsigned long hole_start, hole_end, size;
>  	struct page *pages;
> +	unsigned long count = 0;
>  
>  	pages = NULL;
>  	size = (1 << order) << PAGE_SHIFT;
> @@ -441,6 +442,9 @@ static struct page *kimage_alloc_crash_control_pages(struct kimage *image,
>  	while (hole_end <= crashk_res.end) {
>  		unsigned long i;
>  
> +		if (++count % 256 == 0)
> +			cond_resched();
> +
>  		if (hole_end > KEXEC_CRASH_CONTROL_MEMORY_LIMIT)
>  			break;
>  		/* See if I overlap any of the segments */