[Mesa-dev] [PATCH 1/5/2] i965/fs: Bail out of opt_sampler_eot() if last instruction isn't EOT.
Francisco Jerez
currojerez at riseup.net
Wed Aug 17 22:30:49 UTC 2016
Matt Turner <mattst88 at gmail.com> writes:
> ... instead of assert failing. Can only happen when the program has an
> unconditional infinite loop.
I'm curious how the framebuffer write gets eliminated, I don't think DCE
is smart enough currently to find out that the FB write is unreachable?
> ---
> Sigh.
>
> src/mesa/drivers/dri/i965/brw_fs.cpp | 5 +++--
> 1 file changed, 3 insertions(+), 2 deletions(-)
>
> diff --git a/src/mesa/drivers/dri/i965/brw_fs.cpp b/src/mesa/drivers/dri/i965/brw_fs.cpp
> index d1ac80a..08e9b6c 100644
> --- a/src/mesa/drivers/dri/i965/brw_fs.cpp
> +++ b/src/mesa/drivers/dri/i965/brw_fs.cpp
> @@ -2511,8 +2511,9 @@ fs_visitor::opt_sampler_eot()
> /* Look for a texturing instruction immediately before the final FB_WRITE. */
> bblock_t *block = cfg->blocks[cfg->num_blocks - 1];
> fs_inst *fb_write = (fs_inst *)block->end();
> - assert(fb_write->eot);
> - assert(fb_write->opcode == FS_OPCODE_FB_WRITE_LOGICAL);
> + if (unlikely(!fb_write->eot) ||
> + unlikely(fb_write->opcode != FS_OPCODE_FB_WRITE_LOGICAL))
> + return false;
>
Maybe we want something in between the two? Like:
| if (unlikely(fb_write->opcode != FS_OPCODE_FB_WRITE_LOGICAL))
| return false;
|
| assert(fb_write->eot);
My thinking is that *if* the final instruction of the shader is a
framebuffer write it should have already been marked EOT during
translation.
> /* There wasn't one; nothing to do. */
> if (unlikely(fb_write->prev->is_head_sentinel()))
> --
> 2.7.3
>
> _______________________________________________
> mesa-dev mailing list
> mesa-dev at lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/mesa-dev
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 212 bytes
Desc: not available
URL: <https://lists.freedesktop.org/archives/mesa-dev/attachments/20160817/f18ad16b/attachment.sig>
More information about the mesa-dev
mailing list