On Thu, Sep 15, 2011 at 06:37:33PM -0700, Keith Packard wrote:

I've got this nice patch from Akshay Joshi that removes almost all of the checkpatch.pl warnings from drm/i915. If I don't merge it now, it's going to go stale and be useless; if I merge it only to drm-intel-next, it will be the source of endless conflicts.

However, it's a huge patch (yes, the code was rather sloppy), and doesn't exactly fit into the "critical patches only please" mode of the current stage of 3.1 development.

I've checked the patch very carefully, using the obvious git diff -b to make sure it really doesn't touch anything but whitespace, but also using objdump -s to compare the output of the compiler. There were no differences found with git-diff -b. The only differences found by objdump are two whitespace changes in some debug output messages in intel_bios.c.

I think I have three choices:

1. merge the patch and expect complaints from upstream

2. thank Akshay for his good intentions, discard the patch and hope that he feels motivated enough to do it all over again in time for the 3.2 merge window.

3. thank Akshay for his good intentions and leave the code as-is, forever to ease back-porting of fixes to older kernel versions.

Frankly, if we're ever going to merge whitespace fixups, this would be a pretty darn good time; drm-intel-fixes and drm-intel-next are in-sync as I haven't started pulling 3.2 code into -next.

Well, I've just started to build up patches, so my current queue is tiny ;-)

Otoh massive whitespace changes always annoy when backporting fixes and also when moving forward work-in-progress and old proof-of-concept patches. If you think it's really worth it, go for it. Otherwise I think strictly enforcing checkpatch compliance (and sparse-cleanliness, while we're at it) going forward should get us there pretty quickly (we have some code-churn after all). And then we could fix up the remaining code in 1-2 releases or so - by then that patch should only touch code that's in maintaince mode for older hw, hopefully. -Daniel

On Thu, Sep 15, 2011 at 22:37, Keith Packard keithp@keithp.com wrote:

I've got this nice patch from Akshay Joshi that removes almost all of the checkpatch.pl warnings from drm/i915. If I don't merge it now, it's going to go stale and be useless; if I merge it only to drm-intel-next, it will be the source of endless conflicts.

However, it's a huge patch (yes, the code was rather sloppy), and doesn't exactly fit into the "critical patches only please" mode of the current stage of 3.1 development.

I've checked the patch very carefully, using the obvious git diff -b to make sure it really doesn't touch anything but whitespace, but also using objdump -s to compare the output of the compiler. There were no differences found with git-diff -b. The only differences found by objdump are two whitespace changes in some debug output messages in intel_bios.c.

I think I have three choices:

1. merge the patch and expect complaints from upstream

2. thank Akshay for his good intentions, discard the patch and hope

that he feels motivated enough to do it all over again in time for the 3.2 merge window.

3. thank Akshay for his good intentions and leave the code as-is,

forever to ease back-porting of fixes to older kernel versions.

Frankly, if we're ever going to merge whitespace fixups, this would be a pretty darn good time; drm-intel-fixes and drm-intel-next are in-sync as I haven't started pulling 3.2 code into -next.

I think that if we don´t get to push this patch now, we are unlikely to do it in nearby future. And such kind of cleanup is a nice thing to have.

So I´d vote for option 1, and then 3, in this order.