| Message ID | 1253872613-28746-1-git-send-email-swami.iyer@ti.com (mailing list archive) |
|---|---|
| State | Awaiting Upstream |
| Headers | show |
Hello. Swaminathan S wrote: > Race condition exists between the cppi_interrupt handler and davinci_interrupt > handler w.r.t completing a TX IO. Since DM646x has seperate DMA and USB > Endpoint interrupts cppi_interrupt handler needs to hold the lock while > operating on the endpoint. > Update over previous patch to avoid taking the lock if already taken. Tested > on DM644x, DM355 and DM646x platforms. > Signed-off-by: Swaminathan S <swami.iyer@ti.com> [...] > diff --git a/drivers/usb/musb/cppi_dma.c b/drivers/usb/musb/cppi_dma.c > index c3577bb..cc83ffa 100644 > --- a/drivers/usb/musb/cppi_dma.c > +++ b/drivers/usb/musb/cppi_dma.c > @@ -1154,8 +1154,11 @@ irqreturn_t cppi_interrupt(int irq, void *dev_id) > struct musb_hw_ep *hw_ep = NULL; > u32 rx, tx; > int i, index; > + unsigned long flags; > > cppi = container_of(musb->dma_controller, struct cppi, controller); > + if (cppi->irq) > + spin_lock_irqsave(&musb->lock, flags); > > tibase = musb->ctrl_base; > > @@ -1285,6 +1288,9 @@ irqreturn_t cppi_interrupt(int irq, void *dev_id) > /* write to CPPI EOI register to re-enable interrupts */ > musb_writel(tibase, DAVINCI_CPPI_EOI_REG, 0); > > + if (cppi->irq) > + spin_unlock_irqrestore(&musb->lock, flags); > + > return IRQ_HANDLED; > } > OK, but don't you want to modify davinci.c instead? WBR, Sergei
diff --git a/drivers/usb/musb/cppi_dma.c b/drivers/usb/musb/cppi_dma.c index c3577bb..cc83ffa 100644 --- a/drivers/usb/musb/cppi_dma.c +++ b/drivers/usb/musb/cppi_dma.c @@ -1154,8 +1154,11 @@ irqreturn_t cppi_interrupt(int irq, void *dev_id) struct musb_hw_ep *hw_ep = NULL; u32 rx, tx; int i, index; + unsigned long flags; cppi = container_of(musb->dma_controller, struct cppi, controller); + if (cppi->irq) + spin_lock_irqsave(&musb->lock, flags); tibase = musb->ctrl_base; @@ -1285,6 +1288,9 @@ irqreturn_t cppi_interrupt(int irq, void *dev_id) /* write to CPPI EOI register to re-enable interrupts */ musb_writel(tibase, DAVINCI_CPPI_EOI_REG, 0); + if (cppi->irq) + spin_unlock_irqrestore(&musb->lock, flags); + return IRQ_HANDLED; }
Race condition exists between the cppi_interrupt handler and davinci_interrupt handler w.r.t completing a TX IO. Since DM646x has seperate DMA and USB Endpoint interrupts cppi_interrupt handler needs to hold the lock while operating on the endpoint. Update over previous patch to avoid taking the lock if already taken. Tested on DM644x, DM355 and DM646x platforms. Signed-off-by: Swaminathan S <swami.iyer@ti.com> --- drivers/usb/musb/cppi_dma.c | 6 ++++++ 1 files changed, 6 insertions(+), 0 deletions(-)