[PATCH 00/13] [RFC] IB: PathScale InfiniPath driver

Here is an initial submission from PathScale of a driver for
InfiniPath InfiniBand HCAs. The driver is fairly big -- some single
files are more than the 100 KB limit for lkml posts -- so I've split
it up into a patch series so it can be reviewed inline. The split-up
doesn't make sense functionally but I want to make review as easy as
possible; any final import will merge the driver as a single git
patch.

I've also put the current splitup patchset into my git tree at

git://git.kernel.org/pub/scm/linux/kernel/git/roland/infiniband.git

in the ipath branch.

There are some things I noticed that could maybe be cleaned up, like
having sysctls that set values also settable through module parameters
under /sys/module, code inside #ifndef __KERNEL__ so include files can
be shared with other PathScale code, code in ipath_i2c.c that might be
simplified by using drivers/i2c, etc. I'd like to try to get a sense
of whether I'm being too picky or whether PathScale really does need
to fix these up before the driver is merged.

Basically I'm trying to feel my way as a maintainer so I can find the
right balance between wanting kernel code to be absolutely perfect and
not wanting to put arbitrary hurdles in front of a vendor who has done
a lot of work on contributing an open driver for their hardware.

I am especially interested in feedback about the mergability of this
driver from the broader kernel community, although of course feedback
from the InfiniBand/RDMA community on IB-specific aspects of the
driver is very much appreciated as well.

Thanks,
Roland
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

Relevant Pages

  • Re: [PATCH 00/13] [RFC] IB: PathScale InfiniPath driver
    ... > be shared with other PathScale code, code in ipath_i2c.c that might be ... > to fix these up before the driver is merged. ... please fix this stuff before. ... magnitudes lower than normally expected from your merges. ...
    (Linux-Kernel)
  • Re: [PATCH 0 of 20] [RFC] ipath - PathScale InfiniPath driver
    ... >> But we (the kernel community), don't really accept that as a valid ... As a general rule a driver should push as much functionality to ... the same way by all infiniband drivers. ... interfaces for kernel bypass are not sufficient for flinging packets ...
    (Linux-Kernel)
  • [PATCH 20 of 20] ipath - ethernet emulation driver
    ... The ethernet emulation driver makes an eth* interface available. ... * ipath_ether.c ethernet driver emulation over PathScale Infinipath ... * @lid: the LID ... * @device: the network device ...
    (Linux-Kernel)
  • Re: [SLE] Intel ALC880 RealTek Driver for Linux
    ... Most developers of the kernel community consider non-GPL kernel ... The driver that ships with SuSE 10.0 can be found there... ... The driver that fixed my sound issue on the ThinkPad R30 HD Audio ...
    (SuSE)
  • [PATCH 0 of 18] ipath driver - for inclusion in 2.6.17
    ... This is a submission of the ipath driver for inclusion in 2.6.17. ... We have addressed all earlier rounds of feedback; the driver is stable; ... Miscellaneous fixes requested by Greg and Andrew. ... InfiniPath HT-460, the small form factor InfiniPath HT-460, the InfiniPath ...
    (Linux-Kernel)