lkml.org 
[lkml]   [2013]   [Jan]   [8]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [PATCH 0/2] staging: usbip: refine the lock
On Tue, Jan 08, 2013 at 01:49:00PM +0800, Harvey Yang wrote:
> This patchset refines some spinlocks which maybe not used properly.
>
> [PATCH 1/2]: The function 'usbip_event_add()' may be called in interrupt context on the stub side:
> 'stub_complete'->'stub_enqueue_ret_unlink'->'usbip_event_add'.
> In this function it tries to get the lock 'ud->lock', so we should disable irq when we get this lock in process context.
>
> [PATCH 2/2]: On the client side, we have a virtual hcd driver, there actually no hardware interrupts, so we do not need worry about race conditions caused by irq. To achieve a good performance there is no need to use the interrupt safe spinlock. Just replace them with a non interrupt safe version.

This information needs to be in the patches itself, otherwise it will
never be seen.

Also, why do you think this is going to make anything faster? Have you
measured it? What are the results?

thanks,

greg k-h


\
 
 \ /
  Last update: 2013-01-08 07:41    [W:0.055 / U:1.032 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site