Mail Archive Search: <+subject:/^(?:^\s*(re|sv|fwd|fw)[\[\]\d]*[:>-]+\s*)*Possible\s+race\s+with\s+br_del

Results:

References: [ +subject:/^(?:^\s*(re|sv|fwd|fw)[\[\]\d]*[:>-]+\s*)*Possible\s+race\s+with\s+br_del_if\s*$/: 14 ]

Total 14 documents matching your query.

1. Possible race with br_del_if() (score: 1): Author: >; Date: Thu, 18 Aug 2005 16:40:36 -0500; I've encountered several oops when adding and removing interfaces from bridges while using Xen. Most of the details are available [1]here. The short of it is the following sequence: CPU0 CPU1 add_de; /archives/netdev/2005-08/msg00089.html (9,340 bytes)
2. Re: Possible race with br_del_if() (score: 1): Author: >; Date: Thu, 18 Aug 2005 15:12:02 -0700; Doesn't the mutex in RTNL work right? or are you calling routines with out asserting it?; /archives/netdev/2005-08/msg00090.html (9,363 bytes)
3. Re: Possible race with br_del_if() (score: 1): Author: >; Date: Thu, 18 Aug 2005 17:23:23 -0500; * Stephen Hemminger <shemminger@xxxxxxxx> [2005-08-18 17:11]: unregister_netdevice asserts RTNL, add_del_if() in br_ioctl.c doesn't seem to do so. I don't see it down dev_get_by_index() path either.; /archives/netdev/2005-08/msg00091.html (8,834 bytes)
4. Re: Possible race with br_del_if() (score: 1): Author: >; Date: Thu, 18 Aug 2005 15:35:31 -0700; But the pat to br_dev_ioctl() is via the socket ioctl and that should already have gotten RTNL. dev_ioctl rtnl_lock() dev_ifsioc() dev->do_ioctl --> br_dev_ioctl; /archives/netdev/2005-08/msg00092.html (9,185 bytes)
5. Re: Possible race with br_del_if() (score: 1): Author: >; Date: Thu, 18 Aug 2005 17:56:01 -0500; * Stephen Hemminger <shemminger@xxxxxxxx> [2005-08-18 17:36]: Hrm. OK. It sounds like both paths are doing the right thing w.r.t asserting RTNL, but br_device_event() still gets called with: 1) dev->; /archives/netdev/2005-08/msg00093.html (10,411 bytes)
6. Re: Possible race with br_del_if() (score: 1): Author: >; Date: Fri, 19 Aug 2005 14:10:52 -0500; * Stephen Hemminger <shemminger@xxxxxxxx> [2005-08-18 17:36]: Just to follow-up, the issue was a race between the call_rcu() callback for destroy_nbp() and an unregister_netdev() call. Sometimes the; /archives/netdev/2005-08/msg00098.html (10,508 bytes)
7. Re: Possible race with br_del_if() (score: 1): Author: >; Date: Fri, 19 Aug 2005 12:40:42 -0700; I'll get back to you, this needs some review, I have a bunch of old test suites to dig up for it.; /archives/netdev/2005-08/msg00099.html (10,932 bytes)
8. Possible race with br_del_if() (score: 1): Author: Ryan Harper <ryanh@xxxxxxxxxx>; Date: Thu, 18 Aug 2005 16:40:36 -0500; Hello, I've encountered several oops when adding and removing interfaces from bridges while using Xen. Most of the details are available [1]here. The short of it is the following sequence: CPU0 CPU1; /archives/netdev/2005-08/msg00250.html (9,355 bytes)
9. Re: Possible race with br_del_if() (score: 1): Author: Stephen Hemminger <shemminger@xxxxxxxx>; Date: Thu, 18 Aug 2005 15:12:02 -0700; Doesn't the mutex in RTNL work right? or are you calling routines with out asserting it?; /archives/netdev/2005-08/msg00251.html (9,438 bytes)
10. Re: Possible race with br_del_if() (score: 1): Author: Ryan Harper <ryanh@xxxxxxxxxx>; Date: Thu, 18 Aug 2005 17:23:23 -0500; * Stephen Hemminger <shemminger@xxxxxxxx> [2005-08-18 17:11]: unregister_netdevice asserts RTNL, add_del_if() in br_ioctl.c doesn't seem to do so. I don't see it down dev_get_by_index() path either.; /archives/netdev/2005-08/msg00252.html (8,926 bytes)
11. Re: Possible race with br_del_if() (score: 1): Author: Stephen Hemminger <shemminger@xxxxxxxx>; Date: Thu, 18 Aug 2005 15:35:31 -0700; But the pat to br_dev_ioctl() is via the socket ioctl and that should already have gotten RTNL. dev_ioctl rtnl_lock() dev_ifsioc() dev->do_ioctl --> br_dev_ioctl; /archives/netdev/2005-08/msg00253.html (9,306 bytes)
12. Re: Possible race with br_del_if() (score: 1): Author: Ryan Harper <ryanh@xxxxxxxxxx>; Date: Thu, 18 Aug 2005 17:56:01 -0500; * Stephen Hemminger <shemminger@xxxxxxxx> [2005-08-18 17:36]: Hrm. OK. It sounds like both paths are doing the right thing w.r.t asserting RTNL, but br_device_event() still gets called with: 1) dev->; /archives/netdev/2005-08/msg00254.html (10,564 bytes)
13. Re: Possible race with br_del_if() (score: 1): Author: Ryan Harper <ryanh@xxxxxxxxxx>; Date: Fri, 19 Aug 2005 14:10:52 -0500; * Stephen Hemminger <shemminger@xxxxxxxx> [2005-08-18 17:36]: Just to follow-up, the issue was a race between the call_rcu() callback for destroy_nbp() and an unregister_netdev() call. Sometimes the; /archives/netdev/2005-08/msg00259.html (10,661 bytes)
14. Re: Possible race with br_del_if() (score: 1): Author: Stephen Hemminger <shemminger@xxxxxxxx>; Date: Fri, 19 Aug 2005 12:40:42 -0700; I'll get back to you, this needs some review, I have a bunch of old test suites to dig up for it.; /archives/netdev/2005-08/msg00260.html (11,112 bytes)

Current List: 1 - 14
Page: [1]