jmerkey@wolfmountaingroup.com wrote: [Stefan Richter wrote]... ... Regarding rlock->count, I stand corrected: It works correctly if the debug_lock()...debug_unlock() region can be entered by up to two contexts and if the second one to enter cannot be preempted by the first one. However, since these regions are enclosed in preempt_disable/enable and since the first one to grab the rlock->lock keeps local interrupts disabled until debug_unlock() or even longer, there is always only a single context in the debug_lock()...debug_unlock() region --- which defeats the whole purpose of the rlock_t. Or what am I missing /now/? Independently of this, you cannot use rlock->flags like you currently do. spin_trylock_irqsave would overwrite the flags of CPU A by the flags of CPU B, making the results of spin_unlock_irqrestore in debug_unlock unpredictable. -- Stefan Richter -=====-==--- =--- =-=-= http://arcgraph.de/sr/ --
| Tarkan Erimer | Re: Dual-Licensing Linux Kernel with GPL V2 and GPL V3 |
| WANG Cong | Re: WANTED: kernel projects for CS students |
| Hiten Pandya | Re: up? (emacs docbook xml ide) |
| Rafael J. Wysocki | [Bug #10606] 2.6.26-rc1 regression: ACPI fails to load SDT. - Dell M1530 |
| David Fenyes | sigsetmask()? (LINUX) |
| X X | X11 GIf viewer somewhere? |
| root | Broken pipe when using reboot/halt, etc. |
| Stephen Tweedie | Unmounting root (no kidding!) [was: Some Linux problems---solved] |
git: | |
| Gerrit Renker | [PATCH 27/37] dccp: Integration of dynamic feature activation - part 2 (server side) |
| David Miller | Re: iptables very slow after commit784544739a25c30637397ace5489eeb6e15d7d49 |
| David Miller | [GIT]: Networking |
| Jarek Poplawski | [PATCH] pkt_sched: Destroy gen estimators under rtnl_lock(). |
