On Mon, 1 Jul 2002 23:45:34 -0700
"Greg KH" <[EMAIL PROTECTED]> wrote:
> On Fri, Jun 28, 2002 at 04:47:57PM +0200, Soewono Effendi wrote:
> >
> > And I think there must be some locking to protect
> > if (driver->owner)
> > __MOD_INC_USE_COUNT(driver->owner);
> > and
> >
> > The goal is just to minimize the kernel locking time as much as
> > possible, with the cost of size, I admit that.
>
> Why do you think that the BKL or any lock is needed around the test and
> set functions?
>
> I agree module unloading is extreemly racy, but take a look at the
> current thre
On Fri, Jun 28, 2002 at 04:47:57PM +0200, Soewono Effendi wrote:
>
> And I think there must be some locking to protect
> if (driver->owner)
> __MOD_INC_USE_COUNT(driver->owner);
> and
> if (driver->owner)
> __MOD_DEC_USE_COUNT(driver->owner);
>
> may be:
>
On Thu, Jun 27, 2002 at 09:21:02PM +0200, Oliver Neukum wrote:
>
> PS: Is it true that the IBM guys have a keyboard they make
> a notch into whenever they reduce BKL usage ?
Heh, yes, I think there is something like that in the main BKL
squasher's cubical. I know I enjoy bugging him about it by
About the BKL stuffs...
As I can see on linux-2.5.22 no BKL is needed.
The idea is just to use __MOD_INC_USE_COUNT and __MOD_DEC_USE_COUNT, though I have the
feeling that there should be at least a minimum locking mechanism to guide safe
"driver->owner".
Is this idea also applicable on 2.4 or
Am Donnerstag, 27. Juni 2002 23:41 schrieb David Brownell:
> Oliver Neukum wrote:
> > Am Donnerstag, 27. Juni 2002 21:33 schrieben Sie:
> >>OK, I'll let you fix that BKL stuff (didn't you add it
> >>in the first place? :) but this looks like the locking
> >
> > I did not add it. It was there for a
OK, I'll let you fix that BKL stuff (didn't you add it
in the first place? :) but this looks like the locking
is being done at the wrong level:
> @@ -1349,7 +1351,9 @@
> usbfs_add_device(dev);
>
> /* find drivers willing to handle this device */
> + lock_kernel(); /* guard agai
Am Donnerstag, 27. Juni 2002 20:24 schrieb David Brownell:
> >>>Hotpluggings are rare events ... Besides they take BKL and thus
> >>>couldn't run in concurrency.
> >>
> >>What was the reason they have BKL? In principle it shouldn't
> >>be needed, since all the relevant USB data structures are lo
>>>Hotpluggings are rare events ... Besides they take BKL and thus
>>>couldn't run in concurrency.
>>
>>What was the reason they have BKL? In principle it shouldn't
>>be needed, since all the relevant USB data structures are locked
>>correctly, and there's no comment in the code explaining why
>
Am Donnerstag, 27. Juni 2002 18:20 schrieb David Brownell:
> Oliver Neukum wrote:
> >>after looking around in source codes and searching in mailing list and
> >>G*gle, I still couldn't find any hints why there exists only one
> >> "khubd" in USB Stack.
> >
> > Hotpluggings are rare events ... Bes
On Thu, 27 Jun 2002 09:15:25 -0700
"David Brownell" <[EMAIL PROTECTED]> wrote:
> Periodic urbs get some special treatment, "automagic resubmission" where urbs
> don't get handed back to drivers in most completion callbacks. Ownership
> of the URB there is in a strange "shared between drivers" st
Oliver Neukum wrote:
>>after looking around in source codes and searching in mailing list and
>>G*gle, I still couldn't find any hints why there exists only one "khubd"
>>in USB Stack.
>
> Hotpluggings are rare events ... Besides they take BKL and thus
> couldn't run in concurrency.
What was t
> Well, actually I wanted to know how exactly URB flows through the Linux USB Stack,
>e.g.
> driver allocates URB, submitted to USB Core, goes to HCD, something done there,
> returned to to driver by callback function, etc.
> And how does it apply for all the different transfer modes.
That's th
On Thu, 27 Jun 2002 10:38:37 -0400
"Johannes Erdfelt" <[EMAIL PROTECTED]> wrote:
> Simplicity. [snip]
I see.
>
> > Is there any informations about "URB lifetime" ? who allocates and deallocates it,
> > when one should/shouldn't remove it, etc.
>
> The driver allocates and deallocates the URB
Am Donnerstag, 27. Juni 2002 14:49 schrieb Soewono Effendi:
> Hello there,
>
> after looking around in source codes and searching in mailing list and
> G*gle, I still couldn't find any hints why there exists only one "khubd"
> in USB Stack. Isn't it "better (performance?)" to have one "khubd" for
On Thu, Jun 27, 2002, Soewono Effendi <[EMAIL PROTECTED]> wrote:
> after looking around in source codes and searching in mailing list and G*gle,
> I still couldn't find any hints why there exists only one "khubd" in USB Stack.
> Isn't it "better (performance?)" to have one "khubd" for each HC? (t
16 matches
Mail list logo
