| Message ID | Yuk9vq7t7VhmnOXu@electric-eye.fr.zoreil.com (mailing list archive) |
|---|---|
| State | Accepted |
| Commit | df1c941468fca014ad092f76672966bb412c2848 |
| Delegated to: | Netdev Maintainers |
| Headers | show |
| Series | [v1,net,1/1] net: avoid overflow when rose /proc displays timer information. | expand |
rose /proc code does not serialize timer accesses. Initial report by Bernard F6BVP Pidoux exhibits overflow amounting to 116 ticks on its HZ=250 system. Full timer access serialization would imho be overkill as rose /proc does not enforce consistency between displayed ROSE_STATE_XYZ and timer values during changes of state. The patch may also fix similar behavior in ax25 /proc, ax25 ioctl and netrom /proc as they all exhibit the same timer serialization policy. This point has not been reported though. The sole remaining use of ax25_display_timer - ax25 rtt valuation - may also perform marginally better but I have not analyzed it too deeply. Signed-off-by: Francois Romieu <romieu@fr.zoreil.com> Tested-by: Bernard Pidoux <f6bvp@free.fr> Cc: Thomas DL9SAU Osterried <thomas@osterried.de> Link: https://lore.kernel.org/all/d5e93cc7-a91f-13d3-49a1-b50c11f0f811@free.fr/ --- net/ax25/ax25_timer.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) Bernard, can you formally test and add your "Tested-by:" on this one ? diff --git a/net/ax25/ax25_timer.c b/net/ax25/ax25_timer.c index 85865ebfdfa2..9f7cb0a7c73f 100644 --- a/net/ax25/ax25_timer.c +++ b/net/ax25/ax25_timer.c @@ -108,10 +108,12 @@ int ax25_t1timer_running(ax25_cb *ax25) unsigned long ax25_display_timer(struct timer_list *timer) { + long delta = timer->expires - jiffies; + if (!timer_pending(timer)) return 0; - return timer->expires - jiffies; + return max(0L, delta); } EXPORT_SYMBOL(ax25_display_timer);
Hello: This patch was applied to netdev/net.git (master) by Jakub Kicinski <kuba@kernel.org>: On Tue, 2 Aug 2022 17:07:42 +0200 you wrote: > rose /proc code does not serialize timer accesses. > > Initial report by Bernard F6BVP Pidoux exhibits overflow amounting > to 116 ticks on its HZ=250 system. > > Full timer access serialization would imho be overkill as rose /proc > does not enforce consistency between displayed ROSE_STATE_XYZ and > timer values during changes of state. > > [...] Here is the summary with links: - [v1,net,1/1] net: avoid overflow when rose /proc displays timer information. https://git.kernel.org/netdev/net/c/df1c941468fc You are awesome, thank you!
diff --git a/net/ax25/ax25_timer.c b/net/ax25/ax25_timer.c index 85865ebfdfa2..9f7cb0a7c73f 100644 --- a/net/ax25/ax25_timer.c +++ b/net/ax25/ax25_timer.c @@ -108,10 +108,12 @@ int ax25_t1timer_running(ax25_cb *ax25) unsigned long ax25_display_timer(struct timer_list *timer) { + long delta = timer->expires - jiffies; + if (!timer_pending(timer)) return 0; - return timer->expires - jiffies; + return max(0L, delta); } EXPORT_SYMBOL(ax25_display_timer);
rose /proc code does not serialize timer accesses. Initial report by Bernard F6BVP Pidoux exhibits overflow amounting to 116 ticks on its HZ=250 system. Full timer access serialization would imho be overkill as rose /proc does not enforce consistency between displayed ROSE_STATE_XYZ and timer values during changes of state. The patch may also fix similar behavior in ax25 /proc, ax25 ioctl and netrom /proc as they all exhibit the same timer serialization policy. This point has not been reported though. The sole remaining use of ax25_display_timer - ax25 rtt valuation - may also perform marginally better but I have not analyzed it too deeply. Signed-off-by: Francois Romieu <romieu@fr.zoreil.com> Cc: Thomas DL9SAU Osterried <thomas@osterried.de> Link: https://lore.kernel.org/all/d5e93cc7-a91f-13d3-49a1-b50c11f0f811@free.fr/ --- net/ax25/ax25_timer.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) Bernard, can you formally test and add your "Tested-by:" on this one ?