[openfabrics-ewg] ipath patches for RC3

[Tziporet Koren]

Bryan O'Sullivan wrote:
> I've atttached a tarball of the patches I just generated.
>
> In the fixes directory are fix patches.  In backport are patches to
> backport to a subset of the supported OFED kernels, along with a
> directory called "patches" that contains an annotated series file and
> all of my backport patches, in case it's useful.
>
> For the explicit backport directories, I've at least compile-tested
> against each kernel.  I've also done some runtime tests on 2.6.16, FC4,
> and everything looks good.
>
> This should be in a form that you ought to be able to simply drop in and
> use.  Sorry I haven't been able to cover all of the OFED kernels with
> backport patches, but I've essentially run out of time today.
>
> 	<b
Hi Bryan,

We (Michael and me) have made another effort to merge ipath patches.
Unfortunately the patches still do not apply cleanly in OFED distribution.
So, there's no choice but to ship RC3 as it.
Any critical bug fixes can be added in RC4, but the patch
re-arrangement won't be able to happen in OFED 1.1.

Tziporet

Detailed status (from MST):

- ipath-06-merge-core-verbs.patch fails on apply - apparently was
  created against vanilla kernel and not against OFED which has ehca
  That was trivial, so I resolved it

- Patch patches/ipath-14-rename-products.patch fails.

Applying patch ipath-14-rename-products.patch
patching file drivers/infiniband/hw/ipath/Makefile
Hunk #2 succeeded at 18 with fuzz 2 (offset -1 lines).
patching file drivers/infiniband/hw/ipath/ipath_driver.c
Hunk #2 succeeded at 968 (offset -1 lines).
patching file drivers/infiniband/hw/ipath/ipath_file_ops.c
can't find file to patch at input line 98
Perhaps you used the wrong -p or --strip option?
The text leading up to this was:
--------------------------
|diff --git a/drivers/infiniband/hw/ipath/ipath_ht400.c
b/drivers/infiniband/hw/ipath/ipath_iba6110.c
|rename from drivers/infiniband/hw/ipath/ipath_ht400.c
|rename to drivers/infiniband/hw/ipath/ipath_iba6110.c
|--- a/drivers/infiniband/hw/ipath/ipath_iba6110.c      Thu Aug 24 
13:06:52 2006
-0700
|+++ b/drivers/infiniband/hw/ipath/ipath_iba6110.c      Thu Aug 24 
13:06:55 2006
-0700
--------------------------
No file to patch.  Skipping patch.
14 out of 14 hunks ignored
can't find file to patch at input line 255
Perhaps you used the wrong -p or --strip option?
The text leading up to this was:
--------------------------
|diff --git a/drivers/infiniband/hw/ipath/ipath_pe800.c
b/drivers/infiniband/hw/ipath/ipath_iba6120.c
|rename from drivers/infiniband/hw/ipath/ipath_pe800.c
|rename to drivers/infiniband/hw/ipath/ipath_iba6120.c
|--- a/drivers/infiniband/hw/ipath/ipath_iba6120.c      Thu Aug 24 
13:06:52 2006
-0700
|+++ b/drivers/infiniband/hw/ipath/ipath_iba6120.c      Thu Aug 24 
13:06:55 2006
-0700
--------------------------
No file to patch.  Skipping patch.
16 out of 16 hunks ignored
patching file drivers/infiniband/hw/ipath/ipath_kernel.h
patching file drivers/infiniband/hw/ipath/ipath_registers.h
Patch ipath-14-rename-products.patch does not apply (enforce with -f)

The reason for this seems to be lines like this in patch:

diff --git a/drivers/infiniband/hw/ipath/ipath_ht400.c
b/drivers/infiniband/hw/ipath/ipath_iba6110.c
rename from drivers/infiniband/hw/ipath/ipath_ht400.c
rename to drivers/infiniband/hw/ipath/ipath_iba6110.c
--- a/drivers/infiniband/hw/ipath/ipath_iba6110.c       Thu Aug 24 
13:06:52 2006 -0700
+++ b/drivers/infiniband/hw/ipath/ipath_iba6110.c       Thu Aug 24 
13:06:55 2006 -0700