lkml.org 
[lkml]   [2015]   [Nov]   [9]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [PATCH 03/23] usb: gadget: f_sourcesink: free requests in sourcesink_disable()
On Fri, Nov 06, 2015 at 10:58:39AM +0100, Krzysztof Opasiak wrote:
>
>
> On 11/06/2015 10:48 AM, Peter Chen wrote:
> >On Fri, Nov 06, 2015 at 09:50:11AM +0100, Robert Baldyga wrote:
> >>On 11/06/2015 09:15 AM, Peter Chen wrote:
> >>>On Tue, Nov 03, 2015 at 01:53:42PM +0100, Robert Baldyga wrote:
> >>>>USB requests in SourceSink function are allocated in sourcesink_get_alt()
> >>>>function, so we prefer to free them rather in sourcesink_disable() than
> >>>>in source_sink_complete() when request is completed with error. It provides
> >>>>better symetry in resource management and improves code readability.
> >>>>
> >>>>Signed-off-by: Robert Baldyga <r.baldyga@samsung.com>
> >>>>---
> >>>> drivers/usb/gadget/function/f_sourcesink.c | 33 +++++++++++++++++++++++++++---
> >>>> 1 file changed, 30 insertions(+), 3 deletions(-)
> >>>>
> >>>>diff --git a/drivers/usb/gadget/function/f_sourcesink.c b/drivers/usb/gadget/function/f_sourcesink.c
> >>>>index a8b68c6..d8f5f56 100644
> >>>>--- a/drivers/usb/gadget/function/f_sourcesink.c
> >>>>+++ b/drivers/usb/gadget/function/f_sourcesink.c
> >>>>@@ -22,6 +22,8 @@
> >>>> #include "g_zero.h"
> >>>> #include "u_f.h"
> >>>>
> >>>>+#define REQ_CNT 8
> >>>>+
> >>>
> >>>It would be buffer if we can have module parameter for this like
> >>>qlen at f_loopback.
> >>>>@@ -561,7 +568,6 @@ static void source_sink_complete(struct usb_ep *ep, struct usb_request *req)
> >>>> req->actual, req->length);
> >>>> if (ep == ss->out_ep)
> >>>> check_read_data(ss, req);
> >>>>- free_ep_req(ep, req);
> >>>
> >>>I find the current code may cause memory leak, since above code
> >>>is only called one time.
> >>>
> >>
> >>I don't see what you mean. Can you explain a bit more in which situation
> >>can memory leak take place?
> >
> >Oh, sorry, I have not seen there is a "break;" at the source_sink_start_ep for
> >none-iso endpoint, so, I am wrong, no memory leak here.
> >
> >I have tried changing bulk for 8 requests, the performance improves
> >greatly, but still a little problem for OUT, I will see what's the matter.
> >Besides, it will be better we can have a qlen parameters like f_loopback,
> >in that case, we can improve the performance for gadget, and get the best performance
> >when testing with usbtest, I will do it later.
> >
>
> Moreover, I would suggest to add qlen and iso_qlen params not only
> qlen as even now we are using different qlen for bulk and iso.
>

Krzysztof, would you know why we have different qlen for bulk and iso now?

--

Best Regards,
Peter Chen


\
 
 \ /
  Last update: 2015-11-10 03:41    [W:0.142 / U:0.300 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site