Re: MegaRAID 3008/3108

[Patrick Welche <prlw1%cam.ac.uk@localhost>]

On Wed, Jun 03, 2015 at 07:30:50PM +0000, Christos Zoulas wrote:
> If Frank can post his patches and we can take a look at the OpenBSD
> driver for the I/O performance fix.

I had a go at porting the OpenBSD driver:

  http://www.netbsd.org/~prlw1/0001-From-OpenBSD-with-no-functional-change.patch
  http://www.netbsd.org/~prlw1/0002-WIP-mfii-attaches-with.patch

It attaches:

mfii: "ServeRAID M1215", firmware 24.2.1-0051
scsibus0 at mfii0: 32 targets, 8 luns per target

and seems to do something:

mfii_scsipi_requet
mfii_get_ccb
mfii_scsi_cmd_cdb start XS_CTL_DATA_IN
mfii_load_ccb ccb_len = 36
mfii_scsi_cmd_cdb end
mfii_start begin
mfii_start end

mfii_intr
mfii_my_intr
mfii_postq
mfii_postq before loop
mfii_postq after loop
mfii_postq after bus_dma_map_sync
mfii_postq end

I don't get as far as single user mode, but I don't know where I'm
stuck:  USB keyboard isn't detected (PR 50023?), and no serial port
- hopefully you can spot the obvious hang... Pressing the power
button is detected and various detached messages appear.

I didn't create workqueues - I hope that the OpenBSD linked list is
being processed.

scsipi vs scsi looks rather different.

OpenBSD limits pci_mapreg_map to 8k, but we map 64k.

I took a big KERNEL_LOCK for now just in case my mutex changes aren't right.

Thoughts?

Cheers,

Patrick