forked from luck/tmp_suning_uos_patched
tile: remove #pragma unroll from finv_buffer_remote()
This directive was put in the kernel source before the "pragma unroll" support for tilegx gcc was upstreamed. Remove it for now, and we can put it back later if/when the compiler support is upstreamed. This avoids a warning when building the kernel. This routine is not on a hot path in any case, so the extra optimization here was mostly just for its own sake. Signed-off-by: Chris Metcalf <cmetcalf@mellanox.com>
This commit is contained in:
parent
8e36f722f7
commit
870ee4ff56
|
@ -138,19 +138,13 @@ finv_buffer_remote(void *buffer, size_t size, int hfh)
|
|||
if ((unsigned long)base < (unsigned long)buffer)
|
||||
base = buffer;
|
||||
|
||||
/*
|
||||
* Fire all the loads we need. The MAF only has eight entries
|
||||
* so we can have at most eight outstanding loads, so we
|
||||
* unroll by that amount.
|
||||
*/
|
||||
#pragma unroll 8
|
||||
/* Fire all the loads we need. */
|
||||
for (; p >= base; p -= step_size)
|
||||
force_load(p);
|
||||
|
||||
/*
|
||||
* Repeat, but with finv's instead of loads, to get rid of the
|
||||
* data we just loaded into our own cache and the old home L3.
|
||||
* No need to unroll since finv's don't target a register.
|
||||
* The finv's are guaranteed not to actually flush the data in
|
||||
* the buffer back to their home, since we just read it, so the
|
||||
* lines are clean in cache; we will only invalidate those lines.
|
||||
|
|
Loading…
Reference in New Issue
Block a user