| Message ID | 1364208824-18751-1-git-send-email-coelho@ti.com (mailing list archive) |
|---|---|
| State | Not Applicable, archived |
| Headers | show |
On Mon, Mar 25, 2013 at 12:53:44PM +0200, Luciano Coelho wrote: > Spin locks and completions are expensive in hard IRQ context and cause > problems with RT kernels. In RT kernels, both spin locks and > completions can schedule(), so we can't use them in hard irq context. > > Move handling code into the irq thread function to avoid that. > > Reported-by: Gregoire Gentil <gregoire@alwaysinnovating.com> > Signed-off-by: Luciano Coelho <coelho@ti.com> > --- > drivers/net/wireless/ti/wlcore/main.c | 53 +++++++++++++-------------------- > 1 file changed, 21 insertions(+), 32 deletions(-) > > diff --git a/drivers/net/wireless/ti/wlcore/main.c b/drivers/net/wireless/ti/wlcore/main.c > index 248daa9..c2730a7 100644 > --- a/drivers/net/wireless/ti/wlcore/main.c > +++ b/drivers/net/wireless/ti/wlcore/main.c > @@ -651,6 +651,25 @@ static irqreturn_t wlcore_irq(int irq, void *cookie) > unsigned long flags; > struct wl1271 *wl = cookie; > > + /* complete the ELP completion */ > + spin_lock_irqsave(&wl->wl_lock, flags); > + set_bit(WL1271_FLAG_IRQ_RUNNING, &wl->flags); > + if (wl->elp_compl) { > + complete(wl->elp_compl); > + wl->elp_compl = NULL; > + } > + > + if (test_bit(WL1271_FLAG_SUSPENDED, &wl->flags)) { > + /* don't enqueue a work right now. mark it as pending */ > + set_bit(WL1271_FLAG_PENDING_WORK, &wl->flags); > + wl1271_debug(DEBUG_IRQ, "should not enqueue work"); > + disable_irq_nosync(wl->irq); > + pm_wakeup_event(wl->dev, 0); > + spin_unlock_irqrestore(&wl->wl_lock, flags); > + return IRQ_HANDLED; > + } > + spin_unlock_irqrestore(&wl->wl_lock, flags); I still think _irqrestore() here is wrong, since it will reenable the IRQ line... other than that, it looks alright.
On Mon, 2013-03-25 at 13:11 +0200, Felipe Balbi wrote: > On Mon, Mar 25, 2013 at 12:53:44PM +0200, Luciano Coelho wrote: > > Spin locks and completions are expensive in hard IRQ context and cause > > problems with RT kernels. In RT kernels, both spin locks and > > completions can schedule(), so we can't use them in hard irq context. > > > > Move handling code into the irq thread function to avoid that. > > > > Reported-by: Gregoire Gentil <gregoire@alwaysinnovating.com> > > Signed-off-by: Luciano Coelho <coelho@ti.com> > > --- > > drivers/net/wireless/ti/wlcore/main.c | 53 +++++++++++++-------------------- > > 1 file changed, 21 insertions(+), 32 deletions(-) > > > > diff --git a/drivers/net/wireless/ti/wlcore/main.c b/drivers/net/wireless/ti/wlcore/main.c > > index 248daa9..c2730a7 100644 > > --- a/drivers/net/wireless/ti/wlcore/main.c > > +++ b/drivers/net/wireless/ti/wlcore/main.c > > @@ -651,6 +651,25 @@ static irqreturn_t wlcore_irq(int irq, void *cookie) > > unsigned long flags; > > struct wl1271 *wl = cookie; > > > > + /* complete the ELP completion */ > > + spin_lock_irqsave(&wl->wl_lock, flags); > > + set_bit(WL1271_FLAG_IRQ_RUNNING, &wl->flags); > > + if (wl->elp_compl) { > > + complete(wl->elp_compl); > > + wl->elp_compl = NULL; > > + } > > + > > + if (test_bit(WL1271_FLAG_SUSPENDED, &wl->flags)) { > > + /* don't enqueue a work right now. mark it as pending */ > > + set_bit(WL1271_FLAG_PENDING_WORK, &wl->flags); > > + wl1271_debug(DEBUG_IRQ, "should not enqueue work"); > > + disable_irq_nosync(wl->irq); > > + pm_wakeup_event(wl->dev, 0); > > + spin_unlock_irqrestore(&wl->wl_lock, flags); > > + return IRQ_HANDLED; > > + } > > + spin_unlock_irqrestore(&wl->wl_lock, flags); > > I still think _irqrestore() here is wrong, since it will reenable the > IRQ line... other than that, it looks alright. As we discussed earlier, it won't re-enable, since it was not enabled when we saved. But, in any case, as we agreed, I'll send a separate patch removing it. Don't want to mix two things in the same patch.
On Mon, Mar 25, 2013 at 01:33:25PM +0200, Luciano Coelho wrote: > On Mon, 2013-03-25 at 13:11 +0200, Felipe Balbi wrote: > > On Mon, Mar 25, 2013 at 12:53:44PM +0200, Luciano Coelho wrote: > > > Spin locks and completions are expensive in hard IRQ context and cause > > > problems with RT kernels. In RT kernels, both spin locks and > > > completions can schedule(), so we can't use them in hard irq context. > > > > > > Move handling code into the irq thread function to avoid that. > > > > > > Reported-by: Gregoire Gentil <gregoire@alwaysinnovating.com> > > > Signed-off-by: Luciano Coelho <coelho@ti.com> > > > --- > > > drivers/net/wireless/ti/wlcore/main.c | 53 +++++++++++++-------------------- > > > 1 file changed, 21 insertions(+), 32 deletions(-) > > > > > > diff --git a/drivers/net/wireless/ti/wlcore/main.c b/drivers/net/wireless/ti/wlcore/main.c > > > index 248daa9..c2730a7 100644 > > > --- a/drivers/net/wireless/ti/wlcore/main.c > > > +++ b/drivers/net/wireless/ti/wlcore/main.c > > > @@ -651,6 +651,25 @@ static irqreturn_t wlcore_irq(int irq, void *cookie) > > > unsigned long flags; > > > struct wl1271 *wl = cookie; > > > > > > + /* complete the ELP completion */ > > > + spin_lock_irqsave(&wl->wl_lock, flags); > > > + set_bit(WL1271_FLAG_IRQ_RUNNING, &wl->flags); > > > + if (wl->elp_compl) { > > > + complete(wl->elp_compl); > > > + wl->elp_compl = NULL; > > > + } > > > + > > > + if (test_bit(WL1271_FLAG_SUSPENDED, &wl->flags)) { > > > + /* don't enqueue a work right now. mark it as pending */ > > > + set_bit(WL1271_FLAG_PENDING_WORK, &wl->flags); > > > + wl1271_debug(DEBUG_IRQ, "should not enqueue work"); > > > + disable_irq_nosync(wl->irq); > > > + pm_wakeup_event(wl->dev, 0); > > > + spin_unlock_irqrestore(&wl->wl_lock, flags); > > > + return IRQ_HANDLED; > > > + } > > > + spin_unlock_irqrestore(&wl->wl_lock, flags); > > > > I still think _irqrestore() here is wrong, since it will reenable the > > IRQ line... other than that, it looks alright. > > As we discussed earlier, it won't re-enable, since it was not enabled > when we saved. But, in any case, as we agreed, I'll send a separate > patch removing it. Don't want to mix two things in the same patch. alright... Reviewed-by: Felipe Balbi <balbi@ti.com>
On Mon, 2013-03-25 at 12:53 +0200, Luciano Coelho wrote: > Spin locks and completions are expensive in hard IRQ context and cause > problems with RT kernels. In RT kernels, both spin locks and > completions can schedule(), so we can't use them in hard irq context. > > Move handling code into the irq thread function to avoid that. > > Reported-by: Gregoire Gentil <gregoire@alwaysinnovating.com> > Signed-off-by: Luciano Coelho <coelho@ti.com> > --- Applied.
diff --git a/drivers/net/wireless/ti/wlcore/main.c b/drivers/net/wireless/ti/wlcore/main.c index 248daa9..c2730a7 100644 --- a/drivers/net/wireless/ti/wlcore/main.c +++ b/drivers/net/wireless/ti/wlcore/main.c @@ -651,6 +651,25 @@ static irqreturn_t wlcore_irq(int irq, void *cookie) unsigned long flags; struct wl1271 *wl = cookie; + /* complete the ELP completion */ + spin_lock_irqsave(&wl->wl_lock, flags); + set_bit(WL1271_FLAG_IRQ_RUNNING, &wl->flags); + if (wl->elp_compl) { + complete(wl->elp_compl); + wl->elp_compl = NULL; + } + + if (test_bit(WL1271_FLAG_SUSPENDED, &wl->flags)) { + /* don't enqueue a work right now. mark it as pending */ + set_bit(WL1271_FLAG_PENDING_WORK, &wl->flags); + wl1271_debug(DEBUG_IRQ, "should not enqueue work"); + disable_irq_nosync(wl->irq); + pm_wakeup_event(wl->dev, 0); + spin_unlock_irqrestore(&wl->wl_lock, flags); + return IRQ_HANDLED; + } + spin_unlock_irqrestore(&wl->wl_lock, flags); + /* TX might be handled here, avoid redundant work */ set_bit(WL1271_FLAG_TX_PENDING, &wl->flags); cancel_work_sync(&wl->tx_work); @@ -5998,35 +6017,6 @@ int wlcore_free_hw(struct wl1271 *wl) } EXPORT_SYMBOL_GPL(wlcore_free_hw); -static irqreturn_t wl12xx_hardirq(int irq, void *cookie) -{ - struct wl1271 *wl = cookie; - unsigned long flags; - - wl1271_debug(DEBUG_IRQ, "IRQ"); - - /* complete the ELP completion */ - spin_lock_irqsave(&wl->wl_lock, flags); - set_bit(WL1271_FLAG_IRQ_RUNNING, &wl->flags); - if (wl->elp_compl) { - complete(wl->elp_compl); - wl->elp_compl = NULL; - } - - if (test_bit(WL1271_FLAG_SUSPENDED, &wl->flags)) { - /* don't enqueue a work right now. mark it as pending */ - set_bit(WL1271_FLAG_PENDING_WORK, &wl->flags); - wl1271_debug(DEBUG_IRQ, "should not enqueue work"); - disable_irq_nosync(wl->irq); - pm_wakeup_event(wl->dev, 0); - spin_unlock_irqrestore(&wl->wl_lock, flags); - return IRQ_HANDLED; - } - spin_unlock_irqrestore(&wl->wl_lock, flags); - - return IRQ_WAKE_THREAD; -} - static void wlcore_nvs_cb(const struct firmware *fw, void *context) { struct wl1271 *wl = context; @@ -6068,9 +6058,8 @@ static void wlcore_nvs_cb(const struct firmware *fw, void *context) else irqflags = IRQF_TRIGGER_HIGH | IRQF_ONESHOT; - ret = request_threaded_irq(wl->irq, wl12xx_hardirq, wlcore_irq, - irqflags, - pdev->name, wl); + ret = request_threaded_irq(wl->irq, NULL, wlcore_irq, + irqflags, pdev->name, wl); if (ret < 0) { wl1271_error("request_irq() failed: %d", ret); goto out_free_nvs;
Spin locks and completions are expensive in hard IRQ context and cause problems with RT kernels. In RT kernels, both spin locks and completions can schedule(), so we can't use them in hard irq context. Move handling code into the irq thread function to avoid that. Reported-by: Gregoire Gentil <gregoire@alwaysinnovating.com> Signed-off-by: Luciano Coelho <coelho@ti.com> --- drivers/net/wireless/ti/wlcore/main.c | 53 +++++++++++++-------------------- 1 file changed, 21 insertions(+), 32 deletions(-)