On 4/3/20 1:19 AM, Matthew R. Ochs wrote: > On Thu, Apr 02, 2020 at 05:43:49PM +0200, Frederic Barrat wrote: >> xive is already mapping the trigger page in kernel space and it can be >> accessed through standard APIs, so let's reuse it and simplify the code. >> >> Signed-off-by: Frederic Barrat <fbar...@linux.ibm.com> >> --- >> drivers/scsi/cxlflash/ocxl_hw.c | 17 +++++++---------- >> drivers/scsi/cxlflash/ocxl_hw.h | 1 - >> 2 files changed, 7 insertions(+), 11 deletions(-) >> >> diff --git a/drivers/scsi/cxlflash/ocxl_hw.c >> b/drivers/scsi/cxlflash/ocxl_hw.c >> index 7018cd802569..59452850f71c 100644 >> --- a/drivers/scsi/cxlflash/ocxl_hw.c >> +++ b/drivers/scsi/cxlflash/ocxl_hw.c >> @@ -15,7 +15,8 @@ >> #include <linux/pseudo_fs.h> >> #include <linux/poll.h> >> #include <linux/sched/signal.h> >> - >> +#include <linux/interrupt.h> >> +#include <asm/xive.h> >> #include <misc/ocxl.h> >> >> #include <uapi/misc/cxl.h> >> @@ -180,7 +181,7 @@ static int afu_map_irq(u64 flags, struct >> ocxlflash_context *ctx, int num, >> struct ocxl_hw_afu *afu = ctx->hw_afu; >> struct device *dev = afu->dev; >> struct ocxlflash_irqs *irq; >> - void __iomem *vtrig; >> + struct xive_irq_data *xd; >> u32 virq; >> int rc = 0; >> >> @@ -204,15 +205,14 @@ static int afu_map_irq(u64 flags, struct >> ocxlflash_context *ctx, int num, >> goto err1; >> } >> >> - vtrig = ioremap(irq->ptrig, PAGE_SIZE); >> - if (unlikely(!vtrig)) { >> - dev_err(dev, "%s: Trigger page mapping failed\n", __func__); >> - rc = -ENOMEM; >> + xd = irq_get_handler_data(virq); >> + if (unlikely(!xd)) { >> + dev_err(dev, "%s: Can't get interrupt data\n", __func__); > > The removal of setting the return code injects a bug should this error leg > ever be encountered. So we should either keep the rc statement e.g. -EINVAL, > -ENXIO, -ENODEV, etc., or remove this error leg. I lean towards keeping the > statement.
I like ENXIO for XIVE matters. C. >> goto err2; >> } >> >> irq->virq = virq; >> - irq->vtrig = vtrig; >> + irq->vtrig = xd->trig_mmio; >> out: >>