DPDK patches and discussions
 help / color / mirror / Atom feed
* [PATCH] eal/x86: fix segfaults in waitpkg power intrinsics
@ 2023-11-07 16:19 Bruce Richardson
  2023-11-08  3:19 ` Tyler Retzlaff
  0 siblings, 1 reply; 3+ messages in thread
From: Bruce Richardson @ 2023-11-07 16:19 UTC (permalink / raw)
  To: dev; +Cc: anatoly.burakov, David Hunt, Bruce Richardson

From: David Hunt <david.hunt@intel.com>

The code was recently enhanced to allow the use of the waitpkg
intrinsics rather than the raw assembly in the rte_power functions.
However, the parameters to the intrinsics, while compiling fine, were
incorrect, and would segfault when run on the appropriate hardware.
This patch fixes the intrinsic parameters. Tested on a system with
tpause and umonitor/umwait instructions.

Fixes: 60943c04f3bc ("eal/x86: use intrinsics for power management")

Signed-off-by: David Hunt <david.hunt@intel.com>
Reviewed-by: Bruce Richardson <bruce.richardson@intel.com>
---
 lib/eal/x86/rte_power_intrinsics.c | 14 +++++++-------
 1 file changed, 7 insertions(+), 7 deletions(-)

diff --git a/lib/eal/x86/rte_power_intrinsics.c b/lib/eal/x86/rte_power_intrinsics.c
index 483395dcd5..532a2e646b 100644
--- a/lib/eal/x86/rte_power_intrinsics.c
+++ b/lib/eal/x86/rte_power_intrinsics.c
@@ -40,12 +40,12 @@ static void intel_umonitor(volatile void *addr)
 
 static void intel_umwait(const uint64_t timeout)
 {
+#if defined(RTE_TOOLCHAIN_MSVC) || defined(__WAITPKG__)
+	_umwait(0, timeout);
+#else
 	const uint32_t tsc_l = (uint32_t)timeout;
 	const uint32_t tsc_h = (uint32_t)(timeout >> 32);
 
-#if defined(RTE_TOOLCHAIN_MSVC) || defined(__WAITPKG__)
-	_umwait(tsc_l, tsc_h);
-#else
 	asm volatile(".byte 0xf2, 0x0f, 0xae, 0xf7;"
 			: /* ignore rflags */
 			: "D"(0), /* enter C0.2 */
@@ -208,17 +208,17 @@ rte_power_monitor(const struct rte_power_monitor_cond *pmc,
 int
 rte_power_pause(const uint64_t tsc_timestamp)
 {
-	const uint32_t tsc_l = (uint32_t)tsc_timestamp;
-	const uint32_t tsc_h = (uint32_t)(tsc_timestamp >> 32);
-
 	/* prevent user from running this instruction if it's not supported */
 	if (!wait_supported)
 		return -ENOTSUP;
 
 	/* execute TPAUSE */
 #if defined(RTE_TOOLCHAIN_MSVC) || defined(__WAITPKG__)
-	_tpause(tsc_l, tsc_h);
+	_tpause(0, tsc_timestamp);
 #else
+	const uint32_t tsc_l = (uint32_t)tsc_timestamp;
+	const uint32_t tsc_h = (uint32_t)(tsc_timestamp >> 32);
+
 	asm volatile(".byte 0x66, 0x0f, 0xae, 0xf7;"
 			: /* ignore rflags */
 			: "D"(0), /* enter C0.2 */
-- 
2.39.2


^ permalink raw reply	[flat|nested] 3+ messages in thread

* Re: [PATCH] eal/x86: fix segfaults in waitpkg power intrinsics
  2023-11-07 16:19 [PATCH] eal/x86: fix segfaults in waitpkg power intrinsics Bruce Richardson
@ 2023-11-08  3:19 ` Tyler Retzlaff
  2023-11-08 15:15   ` Thomas Monjalon
  0 siblings, 1 reply; 3+ messages in thread
From: Tyler Retzlaff @ 2023-11-08  3:19 UTC (permalink / raw)
  To: Bruce Richardson; +Cc: dev, anatoly.burakov, David Hunt

On Tue, Nov 07, 2023 at 04:19:01PM +0000, Bruce Richardson wrote:
> From: David Hunt <david.hunt@intel.com>
> 
> The code was recently enhanced to allow the use of the waitpkg
> intrinsics rather than the raw assembly in the rte_power functions.
> However, the parameters to the intrinsics, while compiling fine, were
> incorrect, and would segfault when run on the appropriate hardware.
> This patch fixes the intrinsic parameters. Tested on a system with
> tpause and umonitor/umwait instructions.
> 
> Fixes: 60943c04f3bc ("eal/x86: use intrinsics for power management")
> 
> Signed-off-by: David Hunt <david.hunt@intel.com>
> Reviewed-by: Bruce Richardson <bruce.richardson@intel.com>
> ---

Reviewed-by: Tyler Retzlaff <roretzla@linux.microsoft.com>

^ permalink raw reply	[flat|nested] 3+ messages in thread

* Re: [PATCH] eal/x86: fix segfaults in waitpkg power intrinsics
  2023-11-08  3:19 ` Tyler Retzlaff
@ 2023-11-08 15:15   ` Thomas Monjalon
  0 siblings, 0 replies; 3+ messages in thread
From: Thomas Monjalon @ 2023-11-08 15:15 UTC (permalink / raw)
  To: Bruce Richardson; +Cc: dev, anatoly.burakov, David Hunt, Tyler Retzlaff

08/11/2023 04:19, Tyler Retzlaff:
> On Tue, Nov 07, 2023 at 04:19:01PM +0000, Bruce Richardson wrote:
> > From: David Hunt <david.hunt@intel.com>
> > 
> > The code was recently enhanced to allow the use of the waitpkg
> > intrinsics rather than the raw assembly in the rte_power functions.
> > However, the parameters to the intrinsics, while compiling fine, were
> > incorrect, and would segfault when run on the appropriate hardware.
> > This patch fixes the intrinsic parameters. Tested on a system with
> > tpause and umonitor/umwait instructions.
> > 
> > Fixes: 60943c04f3bc ("eal/x86: use intrinsics for power management")
> > 
> > Signed-off-by: David Hunt <david.hunt@intel.com>
> > Reviewed-by: Bruce Richardson <bruce.richardson@intel.com>
> 
> Reviewed-by: Tyler Retzlaff <roretzla@linux.microsoft.com>

Applied, thanks.




^ permalink raw reply	[flat|nested] 3+ messages in thread

end of thread, other threads:[~2023-11-08 15:15 UTC | newest]

Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2023-11-07 16:19 [PATCH] eal/x86: fix segfaults in waitpkg power intrinsics Bruce Richardson
2023-11-08  3:19 ` Tyler Retzlaff
2023-11-08 15:15   ` 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).