On Tue, Mar 03, 2015 at 10:50:42AM -0500, Benjamin Romer wrote:
> - if (atomic_read(&vdisk->error_count) ==
> - VIRTHBA_ERROR_COUNT) {
> - LOGERR("Throtling SCSICMD errors disk <%d:%d:%d:%llu>\n",
> - scsidev->host->host_no, scsidev->id,
> - scsidev->channel, scsidev->lun);
> - }
> + atomic_read(&vdisk->error_count);
We discussed this earlier and I didn't get back to you, sorry.
atomic_read() is a sinchronization thing, it's not like reading from a
register which may or may not have important side effects. So we can
just delete this line.
> @@ -471,27 +450,21 @@ signalinsert_inner(struct visorchannel *channel, u32 queue, void *msg)
> if (sig_hdr.head == sig_hdr.tail) {
> sig_hdr.num_overflows++;
> if (!SIG_WRITE_FIELD(channel, queue, &sig_hdr, num_overflows))
> - ERRDRV("visor_memregion_write of NumOverflows failed\n");
> -
> - return FALSE;
> + return FALSE;
> }
This is a behavior change. I don't know if it's right or wrong but it
doesn't belong in this patch.
Btw, I'm these patches are quite large. You're going to have to redo
them to fix the three behavior changes. I'm hoping I can just diff the
two patchsets instead of reviewing everything again so please redo them
within the next couple days so that we don't merge other changes in the
mean time. I would like minimal unrelated changes between this patchset
and the next version.
regards,
dan carpenter
_______________________________________________
devel mailing list
devel@xxxxxxxxxxxxxxxxxxxxxx
http://driverdev.linuxdriverproject.org/mailman/listinfo/driverdev-devel
