On Sat, Jun 9, 2012 at 3:56 PM, Hans-Peter Nilsson
wrote:
> Trying --enable-build-with-cxx revealed a build regression for
> cris-elf from the recent atomic support bits. Tested the same,
> no regressions.
>
> Casting those INTVAL's would've been a tiny bit uglier than
> using a temp, thus. Hey,
Trying --enable-build-with-cxx revealed a build regression for
cris-elf from the recent atomic support bits. Tested the same,
no regressions.
Casting those INTVAL's would've been a tiny bit uglier than
using a temp, thus. Hey, those were stashed there as ints
so what's wrong with passing them as
On Thu, Jun 7, 2012 at 9:58 PM, Uros Bizjak wrote:
> Attached patch rewrites x86 soft-FP as an out-of-line function that
> gets conditionally called when exception occurs. This rewrite removes
> 17 instances of the same code from libgcc. In addition, the patch
> unifies a lot of code between 32bi
The second patch that would have hidden the ICE rather than truly fixing it.
At first I was simply annoyed that I didn't have a dump file to look at that
matched up with the actual input to the CSA pass. That alone warranted
splitting out the CLEANUP_CROSSJUMP call to its own pass.
However, I
On 2012-06-09 03:39, Eric Botcazou wrote:
> the 4.7 compiler generates at -O:
>
> load:
> sync
> lwz 10,0(3)
> lwz 11,4(3)
> cmpw 7,10,10
> bne- 7,$+4
> isync
> mr 3,10
> mr 4,11
> blr
>
> Is that really an atomic load?
Nope
One of two patches that make the testcase ICE go away, both
of which make sense, but our release manager thought that
only this simple one was appropriate for the 4.7 branch.
Committed to 4.7 and mainline after testing on i686 with
BOOT_CFLAGS -Os, and x86_64 (for sanity).
r~
PR c++/53602
The shift count was being truncated by the wrong amount.It is now
zero extended from log2 of the mode size rather than the mode size.
This is a (mostly) obvious fix - iant needed a hint to see what was
wrong. The location of this code meant that this would be a very rare
bug: only ti
> The conversion of the __sync post-reload splitters was half
> complete. Since there are nearly no restrictions on what may
> appear between LL and SC, expand all the patterns immediatly.
> This allows significantly easier code generation for subword
> atomic operations.
On PowerPC/Linux, for:
凜孔寊薋 炙 莥鑌噫擈堄 陔府秦 隋 滙冦 埒 憚煴芴犳 但 連僩?
Hi all,
with the priceless support of Tobias I've almost realized the patch
for this PR. In attachment there's the second draft. During the
regression test I have only one error with select_type_4.f90. The
problem is in the destroy_list subroutine when it checks
associated(node) after the first dea
10 matches
Mail list logo
