DPDK patches and discussions
 help / color / mirror / Atom feed
* [dpdk-dev] [PATCH] service: fix shifts to operate on 64 bit integers
@ 2017-07-31 15:58 Harry van Haaren
  2017-07-31 16:17 ` Gaëtan Rivet
  2017-07-31 16:38 ` [dpdk-dev] [PATCH v2] " Harry van Haaren
  0 siblings, 2 replies; 5+ messages in thread
From: Harry van Haaren @ 2017-07-31 15:58 UTC (permalink / raw)
  To: dev; +Cc: Harry van Haaren

This commit fixes shifts to an integer (1 << shift) which
is assumed to be a 32-bit integer. In this case, the shift is
variable and expected to be valid for 64-bit integers. Given that
the expectation to work with 64 bits exists, we must ensure that
the (1 << shift) one in that formula is actually a uin64_t.

Simply defining a const uint64_t and using it ensures the compiler
is aware of the intention. The issue would only manifest if there
were greater than 31 services registered.

Fixes: 21698354c832 ("service: introduce service cores concept")

Signed-off-by: Harry van Haaren <harry.van.haaren@intel.com>
---
 lib/librte_eal/common/rte_service.c | 16 ++++++++++------
 1 file changed, 10 insertions(+), 6 deletions(-)

diff --git a/lib/librte_eal/common/rte_service.c b/lib/librte_eal/common/rte_service.c
index e82b9ad..8c1cffa 100644
--- a/lib/librte_eal/common/rte_service.c
+++ b/lib/librte_eal/common/rte_service.c
@@ -285,8 +285,9 @@ rte_service_unregister(struct rte_service_spec *spec)
 
 	s->internal_flags &= ~(SERVICE_F_REGISTERED);
 
+	const uint64_t one = 1;
 	for (i = 0; i < RTE_MAX_LCORE; i++)
-		lcore_states[i].service_mask &= ~(1 << service_id);
+		lcore_states[i].service_mask &= ~(one << service_id);
 
 	memset(&rte_services[service_id], 0,
 			sizeof(struct rte_service_spec_impl));
@@ -319,6 +320,7 @@ rte_service_runner_func(void *arg)
 {
 	RTE_SET_USED(arg);
 	uint32_t i;
+	const uint64_t one = 1;
 	const int lcore = rte_lcore_id();
 	struct core_state *cs = &lcore_states[lcore];
 
@@ -327,7 +329,7 @@ rte_service_runner_func(void *arg)
 		for (i = 0; i < rte_service_count; i++) {
 			struct rte_service_spec_impl *s = &rte_services[i];
 			if (s->runstate != RUNSTATE_RUNNING ||
-					!(service_mask & (1 << i)))
+					!(service_mask & (one << i)))
 				continue;
 
 			/* check do we need cmpset, if MT safe or <= 1 core
@@ -448,6 +450,7 @@ service_update(struct rte_service_spec *service, uint32_t lcore,
 {
 	uint32_t i;
 	int32_t sid = -1;
+	const uint64_t one = 1;
 
 	for (i = 0; i < RTE_SERVICE_NUM_MAX; i++) {
 		if ((struct rte_service_spec *)&rte_services[i] == service &&
@@ -465,16 +468,16 @@ service_update(struct rte_service_spec *service, uint32_t lcore,
 
 	if (set) {
 		if (*set) {
-			lcore_states[lcore].service_mask |=  (1 << sid);
+			lcore_states[lcore].service_mask |=  (one << sid);
 			rte_services[sid].num_mapped_cores++;
 		} else {
-			lcore_states[lcore].service_mask &= ~(1 << sid);
+			lcore_states[lcore].service_mask &= ~(one << sid);
 			rte_services[sid].num_mapped_cores--;
 		}
 	}
 
 	if (enabled)
-		*enabled = (lcore_states[lcore].service_mask & (1 << sid));
+		*enabled = (lcore_states[lcore].service_mask & (one << sid));
 
 	rte_smp_wmb();
 
@@ -599,6 +602,7 @@ rte_service_lcore_start(uint32_t lcore)
 int32_t
 rte_service_lcore_stop(uint32_t lcore)
 {
+	const uint64_t one = 1;
 	if (lcore >= RTE_MAX_LCORE)
 		return -EINVAL;
 
@@ -607,7 +611,7 @@ rte_service_lcore_stop(uint32_t lcore)
 
 	uint32_t i;
 	for (i = 0; i < RTE_SERVICE_NUM_MAX; i++) {
-		int32_t enabled = lcore_states[i].service_mask & (1 << i);
+		int32_t enabled = lcore_states[i].service_mask & (one << i);
 		int32_t service_running = rte_services[i].runstate !=
 						RUNSTATE_STOPPED;
 		int32_t only_core = rte_services[i].num_mapped_cores == 1;
-- 
2.7.4

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

* Re: [dpdk-dev] [PATCH] service: fix shifts to operate on 64 bit integers
  2017-07-31 15:58 [dpdk-dev] [PATCH] service: fix shifts to operate on 64 bit integers Harry van Haaren
@ 2017-07-31 16:17 ` Gaëtan Rivet
  2017-07-31 16:20   ` Van Haaren, Harry
  2017-07-31 16:38 ` [dpdk-dev] [PATCH v2] " Harry van Haaren
  1 sibling, 1 reply; 5+ messages in thread
From: Gaëtan Rivet @ 2017-07-31 16:17 UTC (permalink / raw)
  To: Harry van Haaren; +Cc: dev

Hi Harry,

On Mon, Jul 31, 2017 at 04:58:27PM +0100, Harry van Haaren wrote:
> This commit fixes shifts to an integer (1 << shift) which
> is assumed to be a 32-bit integer. In this case, the shift is
> variable and expected to be valid for 64-bit integers. Given that
> the expectation to work with 64 bits exists, we must ensure that
> the (1 << shift) one in that formula is actually a uin64_t.
> 
> Simply defining a const uint64_t and using it ensures the compiler
> is aware of the intention. The issue would only manifest if there
> were greater than 31 services registered.
> 
> Fixes: 21698354c832 ("service: introduce service cores concept")
> 
> Signed-off-by: Harry van Haaren <harry.van.haaren@intel.com>
> ---
>  lib/librte_eal/common/rte_service.c | 16 ++++++++++------
>  1 file changed, 10 insertions(+), 6 deletions(-)
> 
> diff --git a/lib/librte_eal/common/rte_service.c b/lib/librte_eal/common/rte_service.c
> index e82b9ad..8c1cffa 100644
> --- a/lib/librte_eal/common/rte_service.c
> +++ b/lib/librte_eal/common/rte_service.c
> @@ -285,8 +285,9 @@ rte_service_unregister(struct rte_service_spec *spec)
>  
>  	s->internal_flags &= ~(SERVICE_F_REGISTERED);
>  
> +	const uint64_t one = 1;
>  	for (i = 0; i < RTE_MAX_LCORE; i++)
> -		lcore_states[i].service_mask &= ~(1 << service_id);
> +		lcore_states[i].service_mask &= ~(one << service_id);

Why not use UINT64_C(1)?

>  
>  	memset(&rte_services[service_id], 0,
>  			sizeof(struct rte_service_spec_impl));
> @@ -319,6 +320,7 @@ rte_service_runner_func(void *arg)
>  {
>  	RTE_SET_USED(arg);
>  	uint32_t i;
> +	const uint64_t one = 1;
>  	const int lcore = rte_lcore_id();
>  	struct core_state *cs = &lcore_states[lcore];
>  
> @@ -327,7 +329,7 @@ rte_service_runner_func(void *arg)
>  		for (i = 0; i < rte_service_count; i++) {
>  			struct rte_service_spec_impl *s = &rte_services[i];
>  			if (s->runstate != RUNSTATE_RUNNING ||
> -					!(service_mask & (1 << i)))
> +					!(service_mask & (one << i)))
>  				continue;
>  
>  			/* check do we need cmpset, if MT safe or <= 1 core
> @@ -448,6 +450,7 @@ service_update(struct rte_service_spec *service, uint32_t lcore,
>  {
>  	uint32_t i;
>  	int32_t sid = -1;
> +	const uint64_t one = 1;
>  
>  	for (i = 0; i < RTE_SERVICE_NUM_MAX; i++) {
>  		if ((struct rte_service_spec *)&rte_services[i] == service &&
> @@ -465,16 +468,16 @@ service_update(struct rte_service_spec *service, uint32_t lcore,
>  
>  	if (set) {
>  		if (*set) {
> -			lcore_states[lcore].service_mask |=  (1 << sid);
> +			lcore_states[lcore].service_mask |=  (one << sid);
>  			rte_services[sid].num_mapped_cores++;
>  		} else {
> -			lcore_states[lcore].service_mask &= ~(1 << sid);
> +			lcore_states[lcore].service_mask &= ~(one << sid);
>  			rte_services[sid].num_mapped_cores--;
>  		}
>  	}
>  
>  	if (enabled)
> -		*enabled = (lcore_states[lcore].service_mask & (1 << sid));
> +		*enabled = (lcore_states[lcore].service_mask & (one << sid));
>  
>  	rte_smp_wmb();
>  
> @@ -599,6 +602,7 @@ rte_service_lcore_start(uint32_t lcore)
>  int32_t
>  rte_service_lcore_stop(uint32_t lcore)
>  {
> +	const uint64_t one = 1;
>  	if (lcore >= RTE_MAX_LCORE)
>  		return -EINVAL;
>  
> @@ -607,7 +611,7 @@ rte_service_lcore_stop(uint32_t lcore)
>  
>  	uint32_t i;
>  	for (i = 0; i < RTE_SERVICE_NUM_MAX; i++) {
> -		int32_t enabled = lcore_states[i].service_mask & (1 << i);
> +		int32_t enabled = lcore_states[i].service_mask & (one << i);
>  		int32_t service_running = rte_services[i].runstate !=
>  						RUNSTATE_STOPPED;
>  		int32_t only_core = rte_services[i].num_mapped_cores == 1;
> -- 
> 2.7.4
> 

-- 
Gaëtan Rivet
6WIND

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

* Re: [dpdk-dev] [PATCH] service: fix shifts to operate on 64 bit integers
  2017-07-31 16:17 ` Gaëtan Rivet
@ 2017-07-31 16:20   ` Van Haaren, Harry
  0 siblings, 0 replies; 5+ messages in thread
From: Van Haaren, Harry @ 2017-07-31 16:20 UTC (permalink / raw)
  To: Gaëtan Rivet; +Cc: dev

> From: Gaëtan Rivet [mailto:gaetan.rivet@6wind.com]
> Sent: Monday, July 31, 2017 5:18 PM
> To: Van Haaren, Harry <harry.van.haaren@intel.com>
> Cc: dev@dpdk.org
> Subject: Re: [dpdk-dev] [PATCH] service: fix shifts to operate on 64 bit integers
> 
> Hi Harry,
> 
> On Mon, Jul 31, 2017 at 04:58:27PM +0100, Harry van Haaren wrote:
> > This commit fixes shifts to an integer (1 << shift) which
> > is assumed to be a 32-bit integer. In this case, the shift is
> > variable and expected to be valid for 64-bit integers. Given that
> > the expectation to work with 64 bits exists, we must ensure that
> > the (1 << shift) one in that formula is actually a uin64_t.
> >
> > Simply defining a const uint64_t and using it ensures the compiler
> > is aware of the intention. The issue would only manifest if there
> > were greater than 31 services registered.
> >
> > Fixes: 21698354c832 ("service: introduce service cores concept")
> >
> > Signed-off-by: Harry van Haaren <harry.van.haaren@intel.com>
> > ---
> >  lib/librte_eal/common/rte_service.c | 16 ++++++++++------
> >  1 file changed, 10 insertions(+), 6 deletions(-)
> >
> > diff --git a/lib/librte_eal/common/rte_service.c b/lib/librte_eal/common/rte_service.c
> > index e82b9ad..8c1cffa 100644
> > --- a/lib/librte_eal/common/rte_service.c
> > +++ b/lib/librte_eal/common/rte_service.c
> > @@ -285,8 +285,9 @@ rte_service_unregister(struct rte_service_spec *spec)
> >
> >  	s->internal_flags &= ~(SERVICE_F_REGISTERED);
> >
> > +	const uint64_t one = 1;
> >  	for (i = 0; i < RTE_MAX_LCORE; i++)
> > -		lcore_states[i].service_mask &= ~(1 << service_id);
> > +		lcore_states[i].service_mask &= ~(one << service_id);
> 
> Why not use UINT64_C(1)?

Mostly because I've never heard of it before :) Thanks for review, still learnin' every day! Sending v2's every other day...

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

* [dpdk-dev] [PATCH v2] service: fix shifts to operate on 64 bit integers
  2017-07-31 15:58 [dpdk-dev] [PATCH] service: fix shifts to operate on 64 bit integers Harry van Haaren
  2017-07-31 16:17 ` Gaëtan Rivet
@ 2017-07-31 16:38 ` Harry van Haaren
  2017-07-31 20:12   ` Thomas Monjalon
  1 sibling, 1 reply; 5+ messages in thread
From: Harry van Haaren @ 2017-07-31 16:38 UTC (permalink / raw)
  To: dev; +Cc: gaetan.rivet, Harry van Haaren

This commit fixes shifts to an integer (1 << shift) which
is assumed to be a 32-bit integer. In this case, the shift is
variable and expected to be valid for 64-bit integers. Given that
the expectation to work with 64 bits exists, we must ensure that
the (1 << shift) one in that formula is actually a uin64_t.

The UINT64_C() macro portably adds the correct suffix to a constant,
informing the compiler that the value is to be assigned 64 bits.

The issue would only manifests when there were greater than 31
services registered.

Fixes: 21698354c832 ("service: introduce service cores concept")

Signed-off-by: Harry van Haaren <harry.van.haaren@intel.com>

---

v2:
- Use UINT64_C() instead of const uint64_t method (Gaetan)
- Refactored to keep checkpatch happy with line-lenghts
---
 lib/librte_eal/common/rte_service.c | 14 ++++++++------
 1 file changed, 8 insertions(+), 6 deletions(-)

diff --git a/lib/librte_eal/common/rte_service.c b/lib/librte_eal/common/rte_service.c
index e82b9ad..7efb76d 100644
--- a/lib/librte_eal/common/rte_service.c
+++ b/lib/librte_eal/common/rte_service.c
@@ -286,7 +286,7 @@ rte_service_unregister(struct rte_service_spec *spec)
 	s->internal_flags &= ~(SERVICE_F_REGISTERED);
 
 	for (i = 0; i < RTE_MAX_LCORE; i++)
-		lcore_states[i].service_mask &= ~(1 << service_id);
+		lcore_states[i].service_mask &= ~(UINT64_C(1) << service_id);
 
 	memset(&rte_services[service_id], 0,
 			sizeof(struct rte_service_spec_impl));
@@ -327,7 +327,7 @@ rte_service_runner_func(void *arg)
 		for (i = 0; i < rte_service_count; i++) {
 			struct rte_service_spec_impl *s = &rte_services[i];
 			if (s->runstate != RUNSTATE_RUNNING ||
-					!(service_mask & (1 << i)))
+					!(service_mask & (UINT64_C(1) << i)))
 				continue;
 
 			/* check do we need cmpset, if MT safe or <= 1 core
@@ -463,18 +463,19 @@ service_update(struct rte_service_spec *service, uint32_t lcore,
 	if (!lcore_states[lcore].is_service_core)
 		return -EINVAL;
 
+	uint64_t sid_mask = UINT64_C(1) << sid;
 	if (set) {
 		if (*set) {
-			lcore_states[lcore].service_mask |=  (1 << sid);
+			lcore_states[lcore].service_mask |= sid_mask;
 			rte_services[sid].num_mapped_cores++;
 		} else {
-			lcore_states[lcore].service_mask &= ~(1 << sid);
+			lcore_states[lcore].service_mask &= ~(sid_mask);
 			rte_services[sid].num_mapped_cores--;
 		}
 	}
 
 	if (enabled)
-		*enabled = (lcore_states[lcore].service_mask & (1 << sid));
+		*enabled = (lcore_states[lcore].service_mask & (sid_mask));
 
 	rte_smp_wmb();
 
@@ -607,7 +608,8 @@ rte_service_lcore_stop(uint32_t lcore)
 
 	uint32_t i;
 	for (i = 0; i < RTE_SERVICE_NUM_MAX; i++) {
-		int32_t enabled = lcore_states[i].service_mask & (1 << i);
+		int32_t enabled =
+			lcore_states[i].service_mask & (UINT64_C(1) << i);
 		int32_t service_running = rte_services[i].runstate !=
 						RUNSTATE_STOPPED;
 		int32_t only_core = rte_services[i].num_mapped_cores == 1;
-- 
2.7.4

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

* Re: [dpdk-dev] [PATCH v2] service: fix shifts to operate on 64 bit integers
  2017-07-31 16:38 ` [dpdk-dev] [PATCH v2] " Harry van Haaren
@ 2017-07-31 20:12   ` Thomas Monjalon
  0 siblings, 0 replies; 5+ messages in thread
From: Thomas Monjalon @ 2017-07-31 20:12 UTC (permalink / raw)
  To: Harry van Haaren; +Cc: dev, gaetan.rivet

31/07/2017 18:38, Harry van Haaren:
> This commit fixes shifts to an integer (1 << shift) which
> is assumed to be a 32-bit integer. In this case, the shift is
> variable and expected to be valid for 64-bit integers. Given that
> the expectation to work with 64 bits exists, we must ensure that
> the (1 << shift) one in that formula is actually a uin64_t.
> 
> The UINT64_C() macro portably adds the correct suffix to a constant,
> informing the compiler that the value is to be assigned 64 bits.
> 
> The issue would only manifests when there were greater than 31
> services registered.
> 
> Fixes: 21698354c832 ("service: introduce service cores concept")
> 
> Signed-off-by: Harry van Haaren <harry.van.haaren@intel.com>

Applied, thanks

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

end of thread, other threads:[~2017-07-31 20:12 UTC | newest]

Thread overview: 5+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2017-07-31 15:58 [dpdk-dev] [PATCH] service: fix shifts to operate on 64 bit integers Harry van Haaren
2017-07-31 16:17 ` Gaëtan Rivet
2017-07-31 16:20   ` Van Haaren, Harry
2017-07-31 16:38 ` [dpdk-dev] [PATCH v2] " Harry van Haaren
2017-07-31 20:12   ` 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).