[MIPS] Skip masking loads and stores via thread pointer.

[MIPS] Skip masking loads and stores via thread pointer.

Register t8 is reserved for thread pointer, and it can not be changed.
Thus, the value it has can be trusted, it points to the safe data zone
and there is no need to guard it.

BUG= https://code.google.com/p/nativeclient/issues/detail?id=2275
TEST= run smoke tests

Committed: https://gerrit.chromium.org/gerrit/gitweb?p=native_client/pnacl-llvm.git;a=commit;h=ce330a6f5f1a6d7786c1ec1fd127926e25c12b82

[Mark Seaborn, 2013-01-15 16:37:03]

LGTM

BTW, on ARM we use "ldr rN, [r9]" for getting the user thread pointer and "ldr
rN, [r9, #4]" for getting the IRT's thread pointer.

This has two advantages:
 * Reading the IRT's thread pointer is as fast as getting the user thread
pointer.
 * We don't need to use a TLS variable in nacl_syscall.S (although we do
currently use one).

The disadvantage is that accessing user TLS variables is slower.

Have you considered doing this for MIPS?

[petarj, 2013-01-22 23:01:38]

On 2013/01/15 16:37:03, Mark Seaborn wrote:
> LGTM
> 
> BTW, on ARM we use "ldr rN, [r9]" for getting the user thread pointer and "ldr
> rN, [r9, #4]" for getting the IRT's thread pointer.
> 
> This has two advantages:
>  * Reading the IRT's thread pointer is as fast as getting the user thread
> pointer.
>  * We don't need to use a TLS variable in nacl_syscall.S (although we do
> currently use one).
> 
> The disadvantage is that accessing user TLS variables is slower.
> 
> Have you considered doing this for MIPS?
Yes, it is on my TODO list, but still not at the top of it.

[petarj, 2013-01-24 18:06:54]

On 2013/01/22 23:01:38, petarj wrote:
> On 2013/01/15 16:37:03, Mark Seaborn wrote:
> > LGTM
> > 
> > BTW, on ARM we use "ldr rN, [r9]" for getting the user thread pointer and
"ldr
> > rN, [r9, #4]" for getting the IRT's thread pointer.
> > 
> > This has two advantages:
> >  * Reading the IRT's thread pointer is as fast as getting the user thread
> > pointer.
> >  * We don't need to use a TLS variable in nacl_syscall.S (although we do
> > currently use one).
> > 
> > The disadvantage is that accessing user TLS variables is slower.
> > 
> > Have you considered doing this for MIPS?
> Yes, it is on my TODO list, but still not at the top of it.

The change has been committed:
https://gerrit.chromium.org/gerrit/gitweb?p=native_client/pnacl-llvm.git;a=co...

[Mark Seaborn, 2013-01-24 18:16:59]

On 24 January 2013 10:06, <petarj@mips.com> wrote:

> The change has been committed:
> https://gerrit.chromium.org/**gerrit/gitweb?p=native_client/**
>
pnacl-llvm.git;a=commit;h=**ce330a6f5f1a6d7786c1ec1fd12792**6e25c12b82<https://gerrit.chromium.org/gerrit/gitweb?p=native_client/pnacl-llvm.git;a=commit;h=ce330a6f5f1a6d7786c1ec1fd127926e25c12b82>
>
>
https://codereview.chromium.**org/11883035/<https://codereview.chromium.org/1...
>

Are you using "git cl push" to commit?  The code review doesn't seem to
have been updated with the gerrit URL above (such as on
https://codereview.chromium.org/11896044), which "git cl push" should do.

Mark

-- 
You received this message because you are subscribed to the Google Groups
"Native-Client-Reviews" group.
To post to this group, send email to native-client-reviews@googlegroups.com.
To unsubscribe from this group, send email to
native-client-reviews+unsubscribe@googlegroups.com.
Visit this group at http://groups.google.com/group/native-client-reviews?hl=en.
For more options, visit https://groups.google.com/groups/opt_out.

[petarj, 2013-01-24 18:40:53]

On 2013/01/24 18:16:59, Mark Seaborn wrote:
> Are you using "git cl push" to commit?  The code review doesn't seem to
> have been updated with the gerrit URL above (such as on
> https://codereview.chromium.org/11896044), which "git cl push" should do.
> 
> Mark

No, I used "git push". I will try "git cl push" next time.

Petar