FW: [PATCH] makedumpfile: Fix a bug when multi-threads feature meets enospace
Masahiko Hayashi
mas-hayashi at tg.jp.nec.com
Fri Feb 9 03:23:37 PST 2018
>-----Original Message-----
>From: Tachibana Masaki(橘 正樹)
>Sent: Friday, February 09, 2018 8:05 PM
>To: Zhou Wenjian <zhouwj.fi at gmail.com>
>Cc: kexec at lists.infradead.org; Hayashi Masahiko(林 正彦) <mas-hayashi at tg.jp.nec.com>
>Subject: RE: [PATCH] makedumpfile: Fix a bug when multi-threads feature meets enospace
>
>Hi Zhou,
>
>Sorry for the late reply.
>Makedumpfile team are busy now for another development.
>I'll reply in about a week.
>
>Thanks
>tachibana
>
>> -----Original Message-----
>> From: kexec [mailto:kexec-bounces at lists.infradead.org] On Behalf Of Zhou Wenjian
>> Sent: Thursday, February 08, 2018 8:53 PM
>> To: kexec at lists.infradead.org
>> Cc: Kumagai Atsushi <ats-kumagai at wm.jp.nec.com>; Suzuki Keiichirou <kei-suzuki at xr.jp.nec.com>
>> Subject: Re: [PATCH] makedumpfile: Fix a bug when multi-threads feature meets enospace
>>
>> ping...
>>
>>
>> On 01/29/2018 08:54 PM, Zhou Wenjian wrote:
>> > Currently, when multi-threads feature meets enospace error,
>> > the main thread will call pthread_join(). However, there is one
>> > thread doing while{} and won't stop.
>> >
>> > pthread_mutex_lock(&info->page_data_mutex);
>> > while (page_data_buf[index].used != FALSE) {
>> > index = (index + 1) % info->num_buffers;
>> > }
>> > page_data_buf[index].used = TRUE;
>> > pthread_mutex_unlock(&info->page_data_mutex);
>> >
>> > Then makedumpfile hangs.
>> >
>> > This patch add a cancel point in while{}.
>> >
>> > Signed-off-by: Zhou Wenjian <zhouwj.fi at gmail.com>
>> > ---
>> > makedumpfile.c | 4 +++-
>> > 1 file changed, 3 insertions(+), 1 deletion(-)
>> >
>> > diff --git a/makedumpfile.c b/makedumpfile.c
>> > index ed138d3..f7ad50c 100644
>> > --- a/makedumpfile.c
>> > +++ b/makedumpfile.c
>> > @@ -7731,11 +7731,13 @@ kdump_thread_function_cyclic(void *arg) {
>> > buf_ready = FALSE;
>> >
>> > pthread_mutex_lock(&info->page_data_mutex);
>> > + pthread_cleanup_push(pthread_mutex_unlock, &info->page_data_mutex);
>> > while (page_data_buf[index].used != FALSE) {
>> > + pthread_testcancel();
>> > index = (index + 1) % info->num_buffers;
>> > }
>> > page_data_buf[index].used = TRUE;
>> > - pthread_mutex_unlock(&info->page_data_mutex);
>> > + pthread_cleanup_pop(1);
>> >
>> > while (buf_ready == FALSE) {
>> > pthread_testcancel();
>>
>> --
>> Thanks
>> Zhou
>>
>>
>> _______________________________________________
>> kexec mailing list
>> kexec at lists.infradead.org
>> http://lists.infradead.org/mailman/listinfo/kexec
More information about the kexec
mailing list