lkml.org 
[lkml]   [2008]   [Oct]   [10]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [git pull] x86 updates for v2.6.28, phase #2 - PAT updates

* Linus Torvalds <torvalds@linux-foundation.org> wrote:

> On Fri, 10 Oct 2008, Ingo Molnar wrote:
> >
> > Ingo Molnar (12):
> > Revert "reduce tlb/cache flush times of agpgart memory allocation"
> > Revert "introduce two APIs for page attribute"
> > Revert "x86: handle error returns in set_memory_*()"
> > Revert "x86: track memtype for RAM in page struct"
> > Revert "x86, cpa: global flush tlb after splitting large page and before doing cpa"
> > Revert "x86, cpa: remove cpa pool code"
> > Revert "x86, cpa: fix taking the pgd_lock with interrupts off"
> > Revert "x86, cpa: dont use large pages for kernel identity mapping with DEBUG_PAGEALLOC"
> > Revert "x86, cpa: make the kernel physical mapping initialization a two pass sequence"
> > Revert "x86, cpa: remove USER permission from the very early identity mapping attribute"
> > Revert "x86, cpa: rename PTE attribute macros for kernel direct mapping in early boot"
> > x86, pat: cleanups
>
> So half of the commits by Suresh were reverted.
>
> Not only that, they were reverted WITH ABSOLUTELY NO EXPLANATIONS OF
> WHY THEY WERE CLEARLY BUGGY PILES OF CRUD. The revert messages are
> just things like
>
> This reverts commit <sha1>.
>
> which makes both the original commit _and_ the revert just totally
> pointless, because we didn't learn anything.

hm, those reverts werent supposed to survive. Again, my bad. I'll clean
it out.

Here is how the screwup happened: a series was sent, i applied it, found
a test failure with it and reported it:

http://www.uwsg.iu.edu/hypermail/linux/kernel/0809.1/2388.html

but to be able to continue testing i temporarily reverted those bits
manually in reverse order, because it took some down to pin down the
breakage and these bits got intermixed with other commits - and i did
not want to rebase commits that came after the broken series.

Then the corrected v2 series arrived (not a delta fix) and i applied
those. The idea was to create a delta patch against the first series so
that i can see the changes.

v2 worked well in testing:

http://www.uwsg.iu.edu/hypermail/linux/kernel/0809.1/2375.html

So it's basically a v1 -> v2 sequence, with artificial reverts surviving
unintentionally. That's why there were no revert messages either: i
never intended them to become public. I even rember having taken a good
look at "git diff 6b5b551..6e3e492", which the v1->v2 delta was.

What i havent done was to squash all these artificial commits together
and create the delta commit - this is one of the few cases where
rebasing of that temporary tail would have been the right thing to do,
before pushing it out.

And i should also have noticed this weird sequence of commit logs when
doing the pull request.

Sorry :-(

Ingo


\
 
 \ /
  Last update: 2008-10-10 18:49    [from the cache]
©2003-2011 Jasper Spaans