Re: [PATCH 2/2] pinctrl/amd: use byte access to clear irq/wake status bits
On Fri, Jul 20, 2018 at 6:38 PM, Daniel Kurtz wrote: > Sounds reasonable. How about: > > - /* Clear interrupt. > -* We must read the pin register again, in case the > -* value was changed while executing > -* generic_handle_irq() above. > + /* > +* Write-1-to-clear irq/wake status bits in MSByte. > +* All other bits in this byte are read-only. > +* This avoids modifying the lower 24-bits > because they may have > +* changed while executing generic_handle_irq() > above. > */ That looks good. Thanks Daniel
Re: [PATCH 2/2] pinctrl/amd: use byte access to clear irq/wake status bits
On Fri, Jul 20, 2018 at 6:38 PM, Daniel Kurtz wrote: > Sounds reasonable. How about: > > - /* Clear interrupt. > -* We must read the pin register again, in case the > -* value was changed while executing > -* generic_handle_irq() above. > + /* > +* Write-1-to-clear irq/wake status bits in MSByte. > +* All other bits in this byte are read-only. > +* This avoids modifying the lower 24-bits > because they may have > +* changed while executing generic_handle_irq() > above. > */ That looks good. Thanks Daniel
Re: [PATCH 2/2] pinctrl/amd: use byte access to clear irq/wake status bits
Hi Daniel, On Tue, Jul 17, 2018 at 6:30 AM Daniel Drake wrote: > > On Mon, Jul 16, 2018 at 7:57 PM, Daniel Kurtz wrote: > > Commit 6afb10267c1692 ("pinctrl/amd: fix masking of GPIO interrupts") > > changed to the clearing of interrupt status bits to a RMW in a critical > > section. This works, but is a bit overkill. > > > > The relevant interrupt/wake status bits are in the Most Significant Byte > > of a 32-bit word. These two are the only write-able bits in this byte. > > I don't have the hardware to test this any more, and I also don't have > any docs to double if those are really the only writable bits, but > looking at the existing driver code it does seem to be the case. > > I think you should retain the comment noting that the value of the > register may have changed since it was read just a few lines above > (and hence explaining more precisely why we make the special effort > just to modify the MSB), just in case there is further rework of this > code in future and we end up walking into the same trap. It was one of > those issues that took a frustratingly long time to figure out... Sounds reasonable. How about: - /* Clear interrupt. -* We must read the pin register again, in case the -* value was changed while executing -* generic_handle_irq() above. + /* +* Write-1-to-clear irq/wake status bits in MSByte. +* All other bits in this byte are read-only. +* This avoids modifying the lower 24-bits because they may have +* changed while executing generic_handle_irq() above. */ > > Thanks > Daniel
Re: [PATCH 2/2] pinctrl/amd: use byte access to clear irq/wake status bits
Hi Daniel, On Tue, Jul 17, 2018 at 6:30 AM Daniel Drake wrote: > > On Mon, Jul 16, 2018 at 7:57 PM, Daniel Kurtz wrote: > > Commit 6afb10267c1692 ("pinctrl/amd: fix masking of GPIO interrupts") > > changed to the clearing of interrupt status bits to a RMW in a critical > > section. This works, but is a bit overkill. > > > > The relevant interrupt/wake status bits are in the Most Significant Byte > > of a 32-bit word. These two are the only write-able bits in this byte. > > I don't have the hardware to test this any more, and I also don't have > any docs to double if those are really the only writable bits, but > looking at the existing driver code it does seem to be the case. > > I think you should retain the comment noting that the value of the > register may have changed since it was read just a few lines above > (and hence explaining more precisely why we make the special effort > just to modify the MSB), just in case there is further rework of this > code in future and we end up walking into the same trap. It was one of > those issues that took a frustratingly long time to figure out... Sounds reasonable. How about: - /* Clear interrupt. -* We must read the pin register again, in case the -* value was changed while executing -* generic_handle_irq() above. + /* +* Write-1-to-clear irq/wake status bits in MSByte. +* All other bits in this byte are read-only. +* This avoids modifying the lower 24-bits because they may have +* changed while executing generic_handle_irq() above. */ > > Thanks > Daniel
Re: [PATCH 2/2] pinctrl/amd: use byte access to clear irq/wake status bits
On Mon, Jul 16, 2018 at 7:57 PM, Daniel Kurtz wrote: > Commit 6afb10267c1692 ("pinctrl/amd: fix masking of GPIO interrupts") > changed to the clearing of interrupt status bits to a RMW in a critical > section. This works, but is a bit overkill. > > The relevant interrupt/wake status bits are in the Most Significant Byte > of a 32-bit word. These two are the only write-able bits in this byte. I don't have the hardware to test this any more, and I also don't have any docs to double if those are really the only writable bits, but looking at the existing driver code it does seem to be the case. I think you should retain the comment noting that the value of the register may have changed since it was read just a few lines above (and hence explaining more precisely why we make the special effort just to modify the MSB), just in case there is further rework of this code in future and we end up walking into the same trap. It was one of those issues that took a frustratingly long time to figure out... Thanks Daniel
Re: [PATCH 2/2] pinctrl/amd: use byte access to clear irq/wake status bits
On Mon, Jul 16, 2018 at 7:57 PM, Daniel Kurtz wrote: > Commit 6afb10267c1692 ("pinctrl/amd: fix masking of GPIO interrupts") > changed to the clearing of interrupt status bits to a RMW in a critical > section. This works, but is a bit overkill. > > The relevant interrupt/wake status bits are in the Most Significant Byte > of a 32-bit word. These two are the only write-able bits in this byte. I don't have the hardware to test this any more, and I also don't have any docs to double if those are really the only writable bits, but looking at the existing driver code it does seem to be the case. I think you should retain the comment noting that the value of the register may have changed since it was read just a few lines above (and hence explaining more precisely why we make the special effort just to modify the MSB), just in case there is further rework of this code in future and we end up walking into the same trap. It was one of those issues that took a frustratingly long time to figure out... Thanks Daniel