[ovs-dev] [PATCH v3] Double postponing to free subtables.

2019-04-23 Thread fuzhantao
From: Zhantao Fu Subtable destruction should be double postponed because readers could still obtain old values while iterating over pvector implementation before its new version published. --- change since v2: - use git send-email to re-send patch change since v1: - delete the unused

[ovs-dev] [PATCH v2] classifier/dpif-netdev: Double postponing to free subtables.

2019-04-22 Thread fuzhantao
From: Zhantao Fu Subtable destruction should be double postponed because readers could still obtain old values while iterating over pvector implementation before its new version published. Changes since v1: - delete the unused parameter in code, modify the subject and commit-message.

[ovs-dev] [PATCH] use double postponing to free subtables.

2019-04-15 Thread fuzhantao
From: Zhantao Fu Now the subtable in classifier/dpcls is postponed by destroy_subtable before move itself out of cls->subtables(pvector).It will raise use-after-free problem. use double postponing to free subtable Signed-off-by: Zhantao Fu --- lib/classifier.c | 31

[ovs-dev] [PATCH] use double postponing to free subtables.

2019-04-15 Thread fuzhantao
From: Zhantao Fu Now the subtable in classifier/dpcls is postponed by destroy_subtable before move itself out of cls->subtables(pvector).It will raise use-after-free problem. use double postponing to free subtable Signed-off-by: Zhantao Fu --- lib/classifier.c | 29

[ovs-dev] [PATCH] use double postponing to free subtables.

2019-04-15 Thread fuzhantao
From: Zhantao Fu Now the subtable in classifier/dpcls is postponed by destroy_subtable before move itself out of cls->subtables(pvector).It will raise use-after-free problem. use double postponing to free subtable Signed-off-by: Zhantao Fu --- lib/classifier.c | 29

Re: [ovs-dev] Questions about rcu_postpone's wrong use in classifier.c

2019-04-02 Thread fuzhantao
--邮件原件- 发件人: Ilya Maximets [mailto:i.maxim...@samsung.com] 发送时间: 2019年4月1日 20:56 收件人: ovs-dev@openvswitch.org; Ben Pfaff ; fuzhantao 主题: Re: [ovs-dev] Questions about rcu_postpone's wrong use in classifier.c Hi. Thanks for the report. It looks like an rcu misusing inside pvector implementatio

Re: [ovs-dev] Questions about rcu_postpone's wrong use in classifier.c

2019-04-01 Thread fuzhantao
msung.com] 发送时间: 2019年4月1日 20:56 收件人: ovs-dev@openvswitch.org; Ben Pfaff ; fuzhantao 主题: Re: [ovs-dev] Questions about rcu_postpone's wrong use in classifier.c Hi. Thanks for the report. It looks like an rcu misusing inside pvector implementation. The following sequence w

[ovs-dev] Questions about rcu_postpone's wrong use in classifier.c

2019-04-01 Thread fuzhantao
Hi all, We find a core dump file in our environment with ovs+dpdk, the following is call stack: #0 0x7f354d6eb237 in raise () from /lib64/libc.so.6 #1 0x7f354d6ec928 in abort () from /lib64/libc.so.6 #2 0x006a6d99 in PAT_abort () #3 0x006a3edd in