Re: [Qemu-devel] [PATCH v5 0/7] tcg: allocate TB structs preceding translate

[Richard Henderson]

On 06/09/2017 12:52 PM, Emilio G. Cota wrote:
> On Thu, Jun 08, 2017 at 22:37:12 -0700, Richard Henderson wrote:
> > This is a follow-up to Emilio's patch set.
> >
> > My primary changes to Emilio's patches are to the first patch, in
> > merging the existing implementations from tcg/ppc/tcg-target.inc.c
> > into util/cacheinfo.c.
> >
> > Then I've a few follow-up patches to take advantage of the new TB
> > placement for arm platforms.  I've had a look at the asm output for
> > ppc64 and s390x, and don't see anything obvious that can be improved.
> >
> > Changes since v4:
> >    * The first patch reorganized a bit for aarch64 and ppc64.
> >      Re-tested on win32, for which there was a Werror.
> >      Incorporated feedback from Emilio re MacOS.
> >    * Fixed the short description for the tcg/arm patches.
>
> This is shaping up quite nicely. Some minor suggestions:
>
> Can we get these checkpatch warnings fixed ..
>
> > === OUTPUT BEGIN ===
> > Checking PATCH 1/7: util: add cacheinfo...
> > ERROR: do not initialise globals to 0 or NULL
> > #149: FILE: util/cacheinfo.c:11:
> > +int qemu_icache_linesize = 0;
> >
> > ERROR: do not initialise globals to 0 or NULL
> > #150: FILE: util/cacheinfo.c:12:
> > +int qemu_dcache_linesize = 0;

These are bogus checkpatch warnings.  If we really want this, we should also 
use -fno-common.  But without that, there is a real difference between 
initialized and non-initialized global variables.

> > ERROR: space prohibited after that '&&' (ctx:ExW)
> > #191: FILE: util/cacheinfo.c:53:
> > +            && buf[i].Cache.Level == 1) {
> >               ^

This is also bogus.  I have no idea what it's attempting to detect.

> .. as well as these?
>
> > Checking PATCH 6/7: tcg/arm: Try pc-relative addresses for movi...
> > ERROR: code indent should never use tabs
> > #54: FILE: tcg/arm/tcg-target.inc.c:446:
> > +^I}$
> >
> > ERROR: code indent should never use tabs
> > #64: FILE: tcg/arm/tcg-target.inc.c:453:
> > +^I}$

Yes, I can fix these.


r~