Date: Thu, 11 Aug 2005 08:25:48 +0000 (UTC) From: Gleb Smirnoff <glebius@FreeBSD.org> To: src-committers@FreeBSD.org, cvs-src@FreeBSD.org, cvs-all@FreeBSD.org Subject: cvs commit: src/sys/netinet if_ether.c Message-ID: <200508110825.j7B8PnTf027445@repoman.freebsd.org>
next in thread | raw e-mail | index | archive | help
glebius 2005-08-11 08:25:48 UTC
FreeBSD src repository
Modified files:
sys/netinet if_ether.c
Log:
o Fix a race between three threads: output path,
incoming ARP packet and route request adding/removing
ARP entries. The root of the problem is that
struct llinfo_arp was accessed without any locks.
To close race we will use locking provided by
rtentry, that references this llinfo_arp:
- Make arplookup() return a locked rtentry.
- In arpresolve() hold the lock provided by
rt_check()/arplookup() until the end of function,
covering all accesses to the rtentry itself and
llinfo_arp it refers to.
- In in_arpinput() do not drop lock provided by
arplookup() during first part of the function.
- Simplify logic in the first part of in_arpinput(),
removing one level of indentation.
- In the second part of in_arpinput() hold rtentry
lock while copying address.
o Fix a condition when route entry is destroyed, while
another thread is contested on its lock:
- When storing a pointer to rtentry in llinfo_arp list,
always add a reference to this rtentry, to prevent
rtentry being destroyed via RTM_DELETE request.
- Remove this reference when removing entry from
llinfo_arp list.
o Further cleanup of arptimer():
- Inline arptfree() into arptimer().
- Use official queue(3) way to pass LIST.
- Hold rtentry lock while reading its structure.
- Do not check that sdl_family is AF_LINK, but
assert this.
Reviewed by: sam
Stress test: http://www.holm.cc/stress/log/cons141.html
Stress test: http://people.freebsd.org/~pho/stress/log/cons144.html
Revision Changes Path
1.139 +183 -157 src/sys/netinet/if_ether.c
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200508110825.j7B8PnTf027445>