code review 6817080: cmd/5g, cmd/6g: pass the full torture test.

cmd/5g, cmd/6g: pass the full torture test.

The patch adds more cases to agenr to allocate registers later,
and makes 6g generate addresses for sgen in something else than
SI and DI. It avoids a complex save/restore sequence that
amounts to allocate a register before descending in subtrees.

Fixes issue 4207.

[remyoudompheng, 2012-11-03 09:05:37]

Hello golang-dev@googlegroups.com (cc: golang-dev@googlegroups.com),

I'd like you to review this change to
https://go.googlecode.com/hg/

[dave_cheney.net, 2012-11-03 11:54:07]

Nice. Thank you.

http://codereview.appspot.com/6817080/diff/7001/src/cmd/5g/gsubr.c
File src/cmd/5g/gsubr.c (right):

http://codereview.appspot.com/6817080/diff/7001/src/cmd/5g/gsubr.c#newcode364
src/cmd/5g/gsubr.c:364: print("registers allocated at\n");
// print

[remyoudompheng, 2012-11-03 12:04:56]

http://codereview.appspot.com/6817080/diff/7001/src/cmd/5g/gsubr.c
File src/cmd/5g/gsubr.c (right):

http://codereview.appspot.com/6817080/diff/7001/src/cmd/5g/gsubr.c#newcode364
src/cmd/5g/gsubr.c:364: print("registers allocated at\n");
The print is added to match 8g. It looks better than bare hexadecimal nonsense
to me, what do you think?

[dave_cheney.net, 2012-11-04 11:34:08]

LGTM.

[dave_cheney.net, 2012-11-09 07:13:41]

On 2012/11/04 11:34:08, dfc wrote:
> LGTM.

gentle ping.

[rsc, 2012-11-12 20:37:10]

LGTM

[remyoudompheng, 2012-11-12 22:56:19]

*** Submitted as http://code.google.com/p/go/source/detail?r=616bfbaf397e ***

cmd/5g, cmd/6g: pass the full torture test.

The patch adds more cases to agenr to allocate registers later,
and makes 6g generate addresses for sgen in something else than
SI and DI. It avoids a complex save/restore sequence that
amounts to allocate a register before descending in subtrees.

Fixes issue 4207.

R=golang-dev, dave, rsc
CC=golang-dev
http://codereview.appspot.com/6817080