| Message ID | c737228e-fde2-33a0-2fdc-d88f0bfd2a9b@cogentembedded.com (mailing list archive) |
|---|---|
| State | Superseded |
| Delegated to: | Geert Uytterhoeven |
| Headers | show |
| Series | [v2] mmc: tmio_mmc_core: don't claim spurious interrupts | expand |
On Sat, Feb 09, 2019 at 10:13:44PM +0300, Sergei Shtylyov wrote: > I have encountered an interrupt storm during the eMMC chip probing (and > the chip finally didn't get detected). It turned out that U-Boot left > the DMAC interrupts enabled while the Linux driver didn't use those. > The SDHI driver's interrupt handler somehow assumes that, even if an > SDIO interrupt didn't happen, it should return IRQ_HANDLED. I think that > if none of the enabled interrupts happened and got handled, we should > return IRQ_NONE -- that way the kernel IRQ code recoginizes a spurious > interrupt and masks it off pretty quickly... > > Signed-off-by: Sergei Shtylyov <sergei.shtylyov@cogentembedded.com> Looks OK in general. > + return ireg ? true : false; return ireg; ?
On 02/10/2019 12:40 AM, Wolfram Sang wrote: >> I have encountered an interrupt storm during the eMMC chip probing (and >> the chip finally didn't get detected). It turned out that U-Boot left >> the DMAC interrupts enabled while the Linux driver didn't use those. >> The SDHI driver's interrupt handler somehow assumes that, even if an >> SDIO interrupt didn't happen, it should return IRQ_HANDLED. I think that >> if none of the enabled interrupts happened and got handled, we should >> return IRQ_NONE -- that way the kernel IRQ code recoginizes a spurious >> interrupt and masks it off pretty quickly... >> >> Signed-off-by: Sergei Shtylyov <sergei.shtylyov@cogentembedded.com> > > Looks OK in general. > >> + return ireg ? true : false; > > return ireg; > > ? How about !!ireg? MBR, Sergei
> >> + return ireg ? true : false; > > > > return ireg; > > > > ? > > How about !!ireg? Better. Still, I wonder how this cast is gonna be different from the cast being done because the return value is declared bool.
On 02/10/2019 08:49 PM, Wolfram Sang wrote: >>>> + return ireg ? true : false; >>> >>> return ireg; >>> >>> ? >> >> How about !!ireg? > > Better. Still, I wonder how this cast is gonna be different from the > cast being done because the return value is declared bool. At least all the variants result in the same (quite bad) code with AArch64 and ARM gcc 4.8.5. Like this one: mov w0, 1 // D.41524, cbnz w23, .L353 // ireg, .L297: mov w0, 0 // D.41524, b .L353 // MBR, Sergei
Index: mmc/drivers/mmc/host/tmio_mmc_core.c =================================================================== --- mmc.orig/drivers/mmc/host/tmio_mmc_core.c +++ mmc/drivers/mmc/host/tmio_mmc_core.c @@ -629,7 +629,7 @@ static bool __tmio_mmc_sdcard_irq(struct return false; } -static void __tmio_mmc_sdio_irq(struct tmio_mmc_host *host) +static bool __tmio_mmc_sdio_irq(struct tmio_mmc_host *host) { struct mmc_host *mmc = host->mmc; struct tmio_mmc_data *pdata = host->pdata; @@ -637,7 +637,7 @@ static void __tmio_mmc_sdio_irq(struct t unsigned int sdio_status; if (!(pdata->flags & TMIO_MMC_SDIO_IRQ)) - return; + return false; status = sd_ctrl_read16(host, CTL_SDIO_STATUS); ireg = status & TMIO_SDIO_MASK_ALL & ~host->sdio_irq_mask; @@ -650,6 +650,8 @@ static void __tmio_mmc_sdio_irq(struct t if (mmc->caps & MMC_CAP_SDIO_IRQ && ireg & TMIO_SDIO_STAT_IOIRQ) mmc_signal_sdio_irq(mmc); + + return ireg ? true : false; } irqreturn_t tmio_mmc_irq(int irq, void *devid) @@ -668,9 +670,10 @@ irqreturn_t tmio_mmc_irq(int irq, void * if (__tmio_mmc_sdcard_irq(host, ireg, status)) return IRQ_HANDLED; - __tmio_mmc_sdio_irq(host); + if (__tmio_mmc_sdio_irq(host)) + return IRQ_HANDLED; - return IRQ_HANDLED; + return IRQ_NONE; } EXPORT_SYMBOL_GPL(tmio_mmc_irq);
I have encountered an interrupt storm during the eMMC chip probing (and the chip finally didn't get detected). It turned out that U-Boot left the DMAC interrupts enabled while the Linux driver didn't use those. The SDHI driver's interrupt handler somehow assumes that, even if an SDIO interrupt didn't happen, it should return IRQ_HANDLED. I think that if none of the enabled interrupts happened and got handled, we should return IRQ_NONE -- that way the kernel IRQ code recoginizes a spurious interrupt and masks it off pretty quickly... Signed-off-by: Sergei Shtylyov <sergei.shtylyov@cogentembedded.com> --- The patch is against Ulf Hansson's 'mmc.git' repo's 'fixes' branch. We seem to have dropped the ball on the version 1... Changes in version 2: - refreshed the patch; - fixed an article and whitespace in the patch description. drivers/mmc/host/tmio_mmc_core.c | 11 +++++++---- 1 file changed, 7 insertions(+), 4 deletions(-)