Oleg Endo wrote:
> 2011-12-21 Oleg Endo
>
> * config/sh/sync.md: Add soft atomics ABI description.
> (fetchop_name): Use 'or' instead of 'ior'.
> (fetchop_insn): Remove.
> (atomic_compare_and_swap_soft):
> Don't insert aligning nop after the write-back instructi
> > Ugh, I've read the middle end code wrongly. Then we can remove
> > fetchop_insn which now becomes to be the same one with fetchop_name.
> > Could you propose the patch with that change and the backslash
> > changes rth suggested? It's pre-approved with those changes.
> >
Attached patch shou
On Sat, 2011-12-17 at 13:32 -0800, Richard Henderson wrote:
> The other thing that should be remembered is that inside a { }
> block you don't need to double-quote all the \'s.
>
Before:
"*
{
return \"\\
mova\\t1f, r0\\n\\
\\t.align\\t2\\n\\
\\tmov\\tr15, r1\\n\\
\\tmov\\t#(0f-1f), r15\\n\\
Oleg Endo wrote:
> If OK, I'd like to add some more verbose comments to sync.md, which also
> describe the ABI that is used for the atomic sequences. Just to have it
> in one place.
Sure. Such comments will be fine.
Regards,
kaz
On Mon, 2011-12-19 at 07:13 +0900, Kaz Kojima wrote:
> You are right. I've misunderstood your changes.
> Sorry for my mess.
Uhm, I could just have left the comments in the patch, but for some
mysterious reasons I deleted them before sending it :T
Sorry for that.
If OK, I'd like to add some more
Oleg Endo wrote:
> I know. That's what I was actually doing. Maybe I should have
> commented it, as it is not so obvious. The aligns are placed in such a
> way, that label 1 will always end up being 4-byte aligned.
You are right. I've misunderstood your changes.
Sorry for my mess.
> Having t
On Sun, 2011-12-18 at 09:15 +0900, Kaz Kojima wrote:
> Your patch doesn't work because SH soft atomic sequences have
> another constraint that label 1 should have been 4-byte aligned.
I know. That's what I was actually doing. Maybe I should have
commented it, as it is not so obvious. The align
Oleg Endo wrote:
> The attached patch should fix the align 2 issues mentioned before and
> also fixes the ior fetchop_name. It should be "or" instead of "ior".
You are right about that nop shouldn't be inserted after
write-back. Thanks for pointing out my thinko.
Your patch doesn't work because
On 12/17/2011 12:13 PM, Oleg Endo wrote:
> @@ -90,13 +90,13 @@
>return \"\\
> mova\\t1f, r0\\n\\
> \\t\\t%2, %4\\n\\
> +\\t.align\\t2\\n\\
The other thing that should be remembered is that inside a { }
block you don't need to double-quote all the \'s.
r~
The attached patch should fix the align 2 issues mentioned before and
also fixes the ior fetchop_name. It should be "or" instead of "ior".
I'm not sure whether it requires re-testing.
Just in case now running:
make -k -check RUNTESTFLAGS="--target_board=sh-sim
\{-m2/-ml/-msoft-atomic,
-m2/-mb/-m
On Mon, 2011-12-05 at 08:07 +0900, Kaz Kojima wrote:
> + return \"\\
> +mova\\t1f, r0\\n\\
> +\\t\\t%2, %4\\n\\
> +\\tmov\\tr15, r1\\n\\
> +\\tmov\\t#(0f-1f), r15\\n\\
> +0:\\tmov.\\t@%1, %0\\n\\
> +\\tcmp/eq\\t%0, %4\\n\\
> +\\tbf\\t1f\\n\\
> +\\tmov.\\t%3, @%1\\n\\
> +\\t.align\\t2\\n\\
> +1:\\
Hi,
The attached patch adds atomic patterns with software atomic
sequences. They are enabled when a new option -msoft-atomic
is specified and the option is default for sh-linux.
Regtested on sh4-unknown-linux-gnu with no new failures and
the doc patch is tested with "make info dvi pdf".
Applied o
12 matches
Mail list logo
