Skip to content

L
linux-davinci
Switch branch/tag
  1. 10 Jun, 2006 1 commit
  3. 08 Jun, 2006 3 commits
    • Matt Reimer's avatar
      [ARM] 3546/1: PATCH: subtle lost interrupts bug on i.MX · e2f04e18
      Matt Reimer authored 
      Patch from Matt Reimer

There is a subtle bug in the GPIO interrupt status register
handling in arch/arm/mach-imx/irq.c:imx_gpio_ack_irq(). The
documentation states that a 1 should be written to the relevant bit to
acknowledge a GPIO interrupt, but that is not what the code does.

The problem is that the |= writes back 1s for all the *other*
interrupts represented in the register, so interrupts could get lost.
For example, if interrupts are pending for GPIO B10 and B12, ISR_B
would have the value 0x00001400. Then when the interrupt code handles
GPIO B10, it eventually calls imx_gpio_ack_irq(IRQ_GPIOB(10)), which
effectively does this:

ISR_B |= 1 << 10;

with the result that (0x00001400 | 0x00000400) is written, clearing
the interrupt status bits for *both* GPIO B10 and B12.

The fix is to write 1s only for the interrupts we want to clear.

The same problem seems to be occurring in the DMA code; this patch
does not address those issues.
Acked-by: default avatarSascha Hauer <s.hauer@pengutronix.de>
Signed-off-by: default avatarMatt Reimer <mreimer@vpop.net>
Signed-off-by: default avatarRussell King <rmk+kernel@arm.linux.org.uk>
      e2f04e18
    • Richard Purdie's avatar
      [ARM] 3547/1: PXA-OHCI: Allow platforms to specify a power budget · 0c27c5d5
      Richard Purdie authored 
      Patch from Richard Purdie

Add a power budget variable to the PXA OHCI platform data and add a
default value for the spitz platform(s) which prevents known failures
with certain USB devices.
Signed-off-by: default avatarRichard Purdie <rpurdie@rpsys.net>
Signed-off-by: default avatarRussell King <rmk+kernel@arm.linux.org.uk>
      0c27c5d5
    • Russell King's avatar
      [ARM] Fix Neponset IRQ handling · d782f33d
      Russell King authored 
      While testing the genirq code on ARM, a condition was found whereby
the Neponset IRQ handler was being re-entered, causing the system
to deadlock.

Under the ARM IRQ code, this would not have been a visible problem
because the "simple" IRQ handling had no re-entrancy protection.

Resolve this by acknowledging the parent interrupt after we mask it
when we are going to handle one of our "special" level-based sources
(from ethernet or USAR chip.)
Signed-off-by: default avatarRussell King <rmk+kernel@arm.linux.org.uk>
      d782f33d
  5. 06 Jun, 2006 1 commit
  7. 05 Jun, 2006 35 commits