DPDK patches and discussions
 help / color / mirror / Atom feed
* [dpdk-dev] [PATCH] fix build on Atom without SSE4 support
@ 2015-12-05  5:00 Mike Sowka
  2015-12-05  5:00 ` [dpdk-dev] [PATCH] sched: " Mike Sowka
  0 siblings, 1 reply; 5+ messages in thread
From: Mike Sowka @ 2015-12-05  5:00 UTC (permalink / raw)
  To: dev; +Cc: Mike Sowka

I'm buiding dpdk on an N2600 Cedarview Atom, using gcc 4.9.2
on Fedora 21. At least one instruction in the optimized version of
rte_sched::grinder_pipe_exists (_mm_testz_si128) requires SSE4.

Mike Sowka (1):
  sched: fix build on Atom without SSE4 support

 lib/librte_sched/rte_sched.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

-- 
2.1.0

^ permalink raw reply	[flat|nested] 5+ messages in thread
* [dpdk-dev] [PATCH] sched: fix build on Atom without SSE4 support
@ 2015-12-05 22:27 Mike Sowka
  2015-12-05 23:47 ` Thomas Monjalon
  0 siblings, 1 reply; 5+ messages in thread
From: Mike Sowka @ 2015-12-05 22:27 UTC (permalink / raw)
  To: dev; +Cc: Mike Sowka

Thanks for the pointers Thomas. Here is a signed-off patch
re-submission with some explanation, to the best of my experience.

Irrelevant of the target, the preprocessor #ifdef SSE2 for the
grinder_pipe_exists function is inadequate since the __mm_testz_si128
function requires SSE4.1, PTEST instruction described in
https://en.wikipedia.org/wiki/SSE4#SSE4.1 (I do no have better spec
reference). I have bumped the preprocessor #ifdef to require SSE4.

The Atom N2600 does not have SSE4, http://ark.intel.com/products/58916,
and so I had trouble building rte_sched with optimized version of
grinder_pipe_exists, with following:
error: inlining failed in call to always_inline _mm_testz_si128’:
       target specific option mismatch

GCC 4.9 correctly identifies my target as not having SSE4, and with
provided patch builds the non-optimized version of grinder_pipe_exists.


Signed-off-by: Mike Sowka <msowka@gmail.com>
---
 lib/librte_sched/rte_sched.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/lib/librte_sched/rte_sched.c b/lib/librte_sched/rte_sched.c
index 21ebf25..6f92aa6 100644
--- a/lib/librte_sched/rte_sched.c
+++ b/lib/librte_sched/rte_sched.c
@@ -55,8 +55,8 @@
 #ifdef RTE_SCHED_VECTOR
 #include <rte_vect.h>
 
-#if defined(__SSE2__)
-#define SCHED_VECTOR_SSE2
+#if defined(__SSE4__)
+#define SCHED_VECTOR_SSE4
 #endif
 
 #endif
@@ -1672,7 +1672,7 @@ grinder_schedule(struct rte_sched_port *port, uint32_t pos)
 	return 1;
 }
 
-#ifdef SCHED_VECTOR_SSE2
+#ifdef SCHED_VECTOR_SSE4
 
 static inline int
 grinder_pipe_exists(struct rte_sched_port *port, uint32_t base_pipe)
-- 
2.1.0

^ permalink raw reply	[flat|nested] 5+ messages in thread

end of thread, other threads:[~2015-12-05 23:48 UTC | newest]

Thread overview: 5+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2015-12-05  5:00 [dpdk-dev] [PATCH] fix build on Atom without SSE4 support Mike Sowka
2015-12-05  5:00 ` [dpdk-dev] [PATCH] sched: " Mike Sowka
2015-12-05 21:07   ` Thomas Monjalon
2015-12-05 22:27 Mike Sowka
2015-12-05 23:47 ` Thomas Monjalon

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).