* [PATCH] app/testpmd: fix lcore ID restriction @ 2024-04-15 19:46 Sivaprasad Tummala 2024-04-15 21:54 ` Stephen Hemminger ` (2 more replies) 0 siblings, 3 replies; 17+ messages in thread From: Sivaprasad Tummala @ 2024-04-15 19:46 UTC (permalink / raw) To: david.marchand, aman.deep.singh, yuying.zhang, ferruh.yigit; +Cc: dev, stable With modern CPUs, it is possible to have higher CPU count thus we can have higher RTE_MAX_LCORES. In testpmd application, the current config forwarding cores option "--nb-cores" is hard limited to 255. The patch fixes this constraint and also adjusts the lcore data structure to 32-bit to align with rte lcore APIs. Fixes: af75078fece3 ("first public release") Cc: stable@dpdk.org Signed-off-by: Sivaprasad Tummala <sivaprasad.tummala@amd.com> --- app/test-pmd/parameters.c | 2 +- app/test-pmd/testpmd.h | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) diff --git a/app/test-pmd/parameters.c b/app/test-pmd/parameters.c index c13f7564bf..6d2bf9b7c1 100644 --- a/app/test-pmd/parameters.c +++ b/app/test-pmd/parameters.c @@ -1072,7 +1072,7 @@ launch_args_parse(int argc, char** argv) case TESTPMD_OPT_NB_CORES_NUM: n = atoi(optarg); if (n > 0 && n <= nb_lcores) - nb_fwd_lcores = (uint8_t) n; + nb_fwd_lcores = (lcoreid_t) n; else rte_exit(EXIT_FAILURE, "nb-cores should be > 0 and <= %d\n", diff --git a/app/test-pmd/testpmd.h b/app/test-pmd/testpmd.h index 0afae7d771..9facd7f281 100644 --- a/app/test-pmd/testpmd.h +++ b/app/test-pmd/testpmd.h @@ -84,7 +84,7 @@ extern volatile uint8_t f_quit; /* Maximum number of pools supported per Rx queue */ #define MAX_MEMPOOL 8 -typedef uint8_t lcoreid_t; +typedef uint32_t lcoreid_t; typedef uint16_t portid_t; typedef uint16_t queueid_t; typedef uint16_t streamid_t; -- 2.40.1 ^ permalink raw reply [flat|nested] 17+ messages in thread
* Re: [PATCH] app/testpmd: fix lcore ID restriction 2024-04-15 19:46 [PATCH] app/testpmd: fix lcore ID restriction Sivaprasad Tummala @ 2024-04-15 21:54 ` Stephen Hemminger 2024-04-15 21:56 ` Stephen Hemminger 2024-04-16 9:55 ` [PATCH v2] " Sivaprasad Tummala 2 siblings, 0 replies; 17+ messages in thread From: Stephen Hemminger @ 2024-04-15 21:54 UTC (permalink / raw) To: Sivaprasad Tummala Cc: david.marchand, aman.deep.singh, yuying.zhang, ferruh.yigit, dev, stable On Mon, 15 Apr 2024 19:46:31 +0000 Sivaprasad Tummala <sivaprasad.tummala@amd.com> wrote: > + nb_fwd_lcores = (lcoreid_t) n; This should really be using strtoul() not atoi() because it offers more error checking. ^ permalink raw reply [flat|nested] 17+ messages in thread
* Re: [PATCH] app/testpmd: fix lcore ID restriction 2024-04-15 19:46 [PATCH] app/testpmd: fix lcore ID restriction Sivaprasad Tummala 2024-04-15 21:54 ` Stephen Hemminger @ 2024-04-15 21:56 ` Stephen Hemminger 2024-04-16 9:55 ` [PATCH v2] " Sivaprasad Tummala 2 siblings, 0 replies; 17+ messages in thread From: Stephen Hemminger @ 2024-04-15 21:56 UTC (permalink / raw) To: Sivaprasad Tummala Cc: david.marchand, aman.deep.singh, yuying.zhang, ferruh.yigit, dev, stable On Mon, 15 Apr 2024 19:46:31 +0000 Sivaprasad Tummala <sivaprasad.tummala@amd.com> wrote: > --- a/app/test-pmd/testpmd.h > +++ b/app/test-pmd/testpmd.h > @@ -84,7 +84,7 @@ extern volatile uint8_t f_quit; > /* Maximum number of pools supported per Rx queue */ > #define MAX_MEMPOOL 8 > > -typedef uint8_t lcoreid_t; > +typedef uint32_t lcoreid_t; > typedef uint16_t portid_t; > typedef uint16_t queueid_t; > typedef uint16_t streamid_t; > -- The other DPDK API's are using unsigned for lcore_id. On the platforms that DPDK supports both are the same size. ^ permalink raw reply [flat|nested] 17+ messages in thread
* [PATCH v2] app/testpmd: fix lcore ID restriction 2024-04-15 19:46 [PATCH] app/testpmd: fix lcore ID restriction Sivaprasad Tummala 2024-04-15 21:54 ` Stephen Hemminger 2024-04-15 21:56 ` Stephen Hemminger @ 2024-04-16 9:55 ` Sivaprasad Tummala 2024-04-19 8:24 ` Ferruh Yigit ` (2 more replies) 2 siblings, 3 replies; 17+ messages in thread From: Sivaprasad Tummala @ 2024-04-16 9:55 UTC (permalink / raw) To: david.marchand, aman.deep.singh, yuying.zhang, ferruh.yigit; +Cc: dev, stable With modern CPUs, it is possible to have higher CPU count thus we can have higher RTE_MAX_LCORES. In testpmd application, the current config forwarding cores option "--nb-cores" is hard limited to 255. The patch fixes this constraint and also adjusts the lcore data structure to 32-bit to align with rte lcore APIs. Fixes: af75078fece3 ("first public release") Cc: stable@dpdk.org Signed-off-by: Sivaprasad Tummala <sivaprasad.tummala@amd.com> --- v2: * Fixed type mistmatch in comparison * Fixed incorrect format specifier --- app/test-pmd/config.c | 6 +++--- app/test-pmd/parameters.c | 4 ++-- app/test-pmd/testpmd.h | 2 +- 3 files changed, 6 insertions(+), 6 deletions(-) diff --git a/app/test-pmd/config.c b/app/test-pmd/config.c index ba1007ace6..6b28c22c96 100644 --- a/app/test-pmd/config.c +++ b/app/test-pmd/config.c @@ -4785,9 +4785,9 @@ fwd_stream_on_other_lcores(uint16_t domain_id, lcoreid_t src_lc, continue; printf("Shared Rx queue group %u queue %hu can't be scheduled on different cores:\n", share_group, share_rxq); - printf(" lcore %hhu Port %hu queue %hu\n", + printf(" lcore %u Port %hu queue %hu\n", src_lc, src_port, src_rxq); - printf(" lcore %hhu Port %hu queue %hu\n", + printf(" lcore %u Port %hu queue %hu\n", lc_id, fs->rx_port, fs->rx_queue); printf("Please use --nb-cores=%hu to limit number of forwarding cores\n", nb_rxq); @@ -5159,7 +5159,7 @@ icmp_echo_config_setup(void) lcoreid_t lc_id; uint16_t sm_id; - if ((nb_txq * nb_fwd_ports) < nb_fwd_lcores) + if ((lcoreid_t)(nb_txq * nb_fwd_ports) < nb_fwd_lcores) cur_fwd_config.nb_fwd_lcores = (lcoreid_t) (nb_txq * nb_fwd_ports); else diff --git a/app/test-pmd/parameters.c b/app/test-pmd/parameters.c index c13f7564bf..22364e09ab 100644 --- a/app/test-pmd/parameters.c +++ b/app/test-pmd/parameters.c @@ -1071,8 +1071,8 @@ launch_args_parse(int argc, char** argv) break; case TESTPMD_OPT_NB_CORES_NUM: n = atoi(optarg); - if (n > 0 && n <= nb_lcores) - nb_fwd_lcores = (uint8_t) n; + if (n > 0 && (lcoreid_t)n <= nb_lcores) + nb_fwd_lcores = (lcoreid_t) n; else rte_exit(EXIT_FAILURE, "nb-cores should be > 0 and <= %d\n", diff --git a/app/test-pmd/testpmd.h b/app/test-pmd/testpmd.h index 0afae7d771..9facd7f281 100644 --- a/app/test-pmd/testpmd.h +++ b/app/test-pmd/testpmd.h @@ -84,7 +84,7 @@ extern volatile uint8_t f_quit; /* Maximum number of pools supported per Rx queue */ #define MAX_MEMPOOL 8 -typedef uint8_t lcoreid_t; +typedef uint32_t lcoreid_t; typedef uint16_t portid_t; typedef uint16_t queueid_t; typedef uint16_t streamid_t; -- 2.34.1 ^ permalink raw reply [flat|nested] 17+ messages in thread
* Re: [PATCH v2] app/testpmd: fix lcore ID restriction 2024-04-16 9:55 ` [PATCH v2] " Sivaprasad Tummala @ 2024-04-19 8:24 ` Ferruh Yigit 2024-04-19 11:30 ` Ferruh Yigit 2024-06-06 11:27 ` [PATCH v3] " Sivaprasad Tummala 2 siblings, 0 replies; 17+ messages in thread From: Ferruh Yigit @ 2024-04-19 8:24 UTC (permalink / raw) To: Sivaprasad Tummala, david.marchand, aman.deep.singh, yuying.zhang Cc: dev, stable On 4/16/2024 10:55 AM, Sivaprasad Tummala wrote: > With modern CPUs, it is possible to have higher > CPU count thus we can have higher RTE_MAX_LCORES. > In testpmd application, the current config forwarding > cores option "--nb-cores" is hard limited to 255. > > The patch fixes this constraint and also adjusts the lcore > data structure to 32-bit to align with rte lcore APIs. > > Fixes: af75078fece3 ("first public release") > Cc: stable@dpdk.org > > Signed-off-by: Sivaprasad Tummala <sivaprasad.tummala@amd.com> > Recheck-request: iol-unit-amd64-testing ^ permalink raw reply [flat|nested] 17+ messages in thread
* Re: [PATCH v2] app/testpmd: fix lcore ID restriction 2024-04-16 9:55 ` [PATCH v2] " Sivaprasad Tummala 2024-04-19 8:24 ` Ferruh Yigit @ 2024-04-19 11:30 ` Ferruh Yigit 2024-06-06 9:56 ` Tummala, Sivaprasad 2024-06-13 16:51 ` Ferruh Yigit 2024-06-06 11:27 ` [PATCH v3] " Sivaprasad Tummala 2 siblings, 2 replies; 17+ messages in thread From: Ferruh Yigit @ 2024-04-19 11:30 UTC (permalink / raw) To: Sivaprasad Tummala, david.marchand, aman.deep.singh, yuying.zhang Cc: dev, stable On 4/16/2024 10:55 AM, Sivaprasad Tummala wrote: > diff --git a/app/test-pmd/config.c b/app/test-pmd/config.c > index ba1007ace6..6b28c22c96 100644 > --- a/app/test-pmd/config.c > +++ b/app/test-pmd/config.c > @@ -4785,9 +4785,9 @@ fwd_stream_on_other_lcores(uint16_t domain_id, lcoreid_t src_lc, > continue; > printf("Shared Rx queue group %u queue %hu can't be scheduled on different cores:\n", > share_group, share_rxq); > - printf(" lcore %hhu Port %hu queue %hu\n", > + printf(" lcore %u Port %hu queue %hu\n", > src_lc, src_port, src_rxq); > - printf(" lcore %hhu Port %hu queue %hu\n", > + printf(" lcore %u Port %hu queue %hu\n", > lc_id, fs->rx_port, fs->rx_queue); > printf("Please use --nb-cores=%hu to limit number of forwarding cores\n", > nb_rxq); > @@ -5159,7 +5159,7 @@ icmp_echo_config_setup(void) > lcoreid_t lc_id; > uint16_t sm_id; > > - if ((nb_txq * nb_fwd_ports) < nb_fwd_lcores) > + if ((lcoreid_t)(nb_txq * nb_fwd_ports) < nb_fwd_lcores) > cur_fwd_config.nb_fwd_lcores = (lcoreid_t) > (nb_txq * nb_fwd_ports); > Hi Sivaprasad, Is this '(lcoreid_t)' cast required? Because of integer promotion I think result will be correct without casting. (And without integer promotion considered, casting needs to be done on one of the variables, not to the result, because result may be already cast down I think. Anyway this is not required for this case since variables are u16.) ^ permalink raw reply [flat|nested] 17+ messages in thread
* RE: [PATCH v2] app/testpmd: fix lcore ID restriction 2024-04-19 11:30 ` Ferruh Yigit @ 2024-06-06 9:56 ` Tummala, Sivaprasad 2024-06-13 16:51 ` Ferruh Yigit 1 sibling, 0 replies; 17+ messages in thread From: Tummala, Sivaprasad @ 2024-06-06 9:56 UTC (permalink / raw) To: Yigit, Ferruh, david.marchand, aman.deep.singh, yuying.zhang; +Cc: dev, stable [AMD Official Use Only - AMD Internal Distribution Only] Hi Ferruh, > -----Original Message----- > From: Yigit, Ferruh <Ferruh.Yigit@amd.com> > Sent: Friday, April 19, 2024 5:00 PM > To: Tummala, Sivaprasad <Sivaprasad.Tummala@amd.com>; > david.marchand@redhat.com; aman.deep.singh@intel.com; > yuying.zhang@intel.com > Cc: dev@dpdk.org; stable@dpdk.org > Subject: Re: [PATCH v2] app/testpmd: fix lcore ID restriction > > On 4/16/2024 10:55 AM, Sivaprasad Tummala wrote: > > diff --git a/app/test-pmd/config.c b/app/test-pmd/config.c index > > ba1007ace6..6b28c22c96 100644 > > --- a/app/test-pmd/config.c > > +++ b/app/test-pmd/config.c > > @@ -4785,9 +4785,9 @@ fwd_stream_on_other_lcores(uint16_t domain_id, > lcoreid_t src_lc, > > continue; > > printf("Shared Rx queue group %u queue %hu can't be > scheduled on different cores:\n", > > share_group, share_rxq); > > - printf(" lcore %hhu Port %hu queue %hu\n", > > + printf(" lcore %u Port %hu queue %hu\n", > > src_lc, src_port, src_rxq); > > - printf(" lcore %hhu Port %hu queue %hu\n", > > + printf(" lcore %u Port %hu queue %hu\n", > > lc_id, fs->rx_port, fs->rx_queue); > > printf("Please use --nb-cores=%hu to limit number of > forwarding cores\n", > > nb_rxq); > > @@ -5159,7 +5159,7 @@ icmp_echo_config_setup(void) > > lcoreid_t lc_id; > > uint16_t sm_id; > > > > - if ((nb_txq * nb_fwd_ports) < nb_fwd_lcores) > > + if ((lcoreid_t)(nb_txq * nb_fwd_ports) < nb_fwd_lcores) > > cur_fwd_config.nb_fwd_lcores = (lcoreid_t) > > (nb_txq * nb_fwd_ports); > > > > Hi Sivaprasad, > > Is this '(lcoreid_t)' cast required? Because of integer promotion I think result will > be correct without casting. > > (And without integer promotion considered, casting needs to be done on one of > the variables, not to the result, because result may be already cast down I think. > Anyway this is not required for this case since variables are u16.) Agreed., the change was added to address a compilation issue on Centos7. It seems The compiler was reporting a false alarm and I will revert this change in the next patch. ^ permalink raw reply [flat|nested] 17+ messages in thread
* Re: [PATCH v2] app/testpmd: fix lcore ID restriction 2024-04-19 11:30 ` Ferruh Yigit 2024-06-06 9:56 ` Tummala, Sivaprasad @ 2024-06-13 16:51 ` Ferruh Yigit 2024-06-13 19:13 ` Stephen Hemminger 1 sibling, 1 reply; 17+ messages in thread From: Ferruh Yigit @ 2024-06-13 16:51 UTC (permalink / raw) To: Sivaprasad Tummala, david.marchand, aman.deep.singh, yuying.zhang Cc: dev, stable On 4/19/2024 12:30 PM, Ferruh Yigit wrote: > On 4/16/2024 10:55 AM, Sivaprasad Tummala wrote: >> diff --git a/app/test-pmd/config.c b/app/test-pmd/config.c >> index ba1007ace6..6b28c22c96 100644 >> --- a/app/test-pmd/config.c >> +++ b/app/test-pmd/config.c >> @@ -4785,9 +4785,9 @@ fwd_stream_on_other_lcores(uint16_t domain_id, lcoreid_t src_lc, >> continue; >> printf("Shared Rx queue group %u queue %hu can't be scheduled on different cores:\n", >> share_group, share_rxq); >> - printf(" lcore %hhu Port %hu queue %hu\n", >> + printf(" lcore %u Port %hu queue %hu\n", >> src_lc, src_port, src_rxq); >> - printf(" lcore %hhu Port %hu queue %hu\n", >> + printf(" lcore %u Port %hu queue %hu\n", >> lc_id, fs->rx_port, fs->rx_queue); >> printf("Please use --nb-cores=%hu to limit number of forwarding cores\n", >> nb_rxq); >> @@ -5159,7 +5159,7 @@ icmp_echo_config_setup(void) >> lcoreid_t lc_id; >> uint16_t sm_id; >> >> - if ((nb_txq * nb_fwd_ports) < nb_fwd_lcores) >> + if ((lcoreid_t)(nb_txq * nb_fwd_ports) < nb_fwd_lcores) >> cur_fwd_config.nb_fwd_lcores = (lcoreid_t) >> (nb_txq * nb_fwd_ports); >> > > Hi Sivaprasad, > > Is this '(lcoreid_t)' cast required? Because of integer promotion I > think result will be correct without casting. > > (And without integer promotion considered, casting needs to be done on > one of the variables, not to the result, because result may be already > cast down I think. Anyway this is not required for this case since > variables are u16.) > Why casing required (for record) is, 'nb_txq' -> uint16_t, promoted to 'int' 'nb_fwd_ports' -> uint16_t, promoted to 'int' (nb_txq * nb_fwd_ports) -> result 'int' nb_fwd_lcores -> 'uint32_t' comparison between 'int' & 'uint32_t' gives warning. After some compiler version it is smart enough to not give a warning, but casting is required for old compilers. And back to my comment above, casting one of the parameter to 'lcoreid_t' also works, as it forcing promotion to 'unsigned int' instead. But logically casting looks odd, so keeping casting result. ^ permalink raw reply [flat|nested] 17+ messages in thread
* Re: [PATCH v2] app/testpmd: fix lcore ID restriction 2024-06-13 16:51 ` Ferruh Yigit @ 2024-06-13 19:13 ` Stephen Hemminger 2024-06-14 9:01 ` Ferruh Yigit 0 siblings, 1 reply; 17+ messages in thread From: Stephen Hemminger @ 2024-06-13 19:13 UTC (permalink / raw) To: Ferruh Yigit Cc: Sivaprasad Tummala, david.marchand, aman.deep.singh, yuying.zhang, dev, stable On Thu, 13 Jun 2024 17:51:14 +0100 Ferruh Yigit <ferruh.yigit@amd.com> wrote: > > Hi Sivaprasad, > > > > Is this '(lcoreid_t)' cast required? Because of integer promotion I > > think result will be correct without casting. > > > > (And without integer promotion considered, casting needs to be done on > > one of the variables, not to the result, because result may be already > > cast down I think. Anyway this is not required for this case since > > variables are u16.) > > > > Why casing required (for record) is, > 'nb_txq' -> uint16_t, promoted to 'int' > 'nb_fwd_ports' -> uint16_t, promoted to 'int' > (nb_txq * nb_fwd_ports) -> result 'int' > nb_fwd_lcores -> 'uint32_t' > > comparison between 'int' & 'uint32_t' gives warning. After some compiler > version it is smart enough to not give a warning, but casting is > required for old compilers. > > And back to my comment above, casting one of the parameter to > 'lcoreid_t' also works, as it forcing promotion to 'unsigned int' > instead. But logically casting looks odd, so keeping casting result. Where is the integer promotion happening? Also would be better to use strtoul() instead of atoi() when parsing the args. That gives better error handling and handles someone passing negative number correctly. ^ permalink raw reply [flat|nested] 17+ messages in thread
* Re: [PATCH v2] app/testpmd: fix lcore ID restriction 2024-06-13 19:13 ` Stephen Hemminger @ 2024-06-14 9:01 ` Ferruh Yigit 2024-06-14 15:27 ` Stephen Hemminger 0 siblings, 1 reply; 17+ messages in thread From: Ferruh Yigit @ 2024-06-14 9:01 UTC (permalink / raw) To: Stephen Hemminger Cc: Sivaprasad Tummala, david.marchand, aman.deep.singh, yuying.zhang, dev, stable On 6/13/2024 8:13 PM, Stephen Hemminger wrote: > On Thu, 13 Jun 2024 17:51:14 +0100 > Ferruh Yigit <ferruh.yigit@amd.com> wrote: > >>> Hi Sivaprasad, >>> >>> Is this '(lcoreid_t)' cast required? Because of integer promotion I >>> think result will be correct without casting. >>> >>> (And without integer promotion considered, casting needs to be done on >>> one of the variables, not to the result, because result may be already >>> cast down I think. Anyway this is not required for this case since >>> variables are u16.) >>> >> >> Why casing required (for record) is, >> 'nb_txq' -> uint16_t, promoted to 'int' >> 'nb_fwd_ports' -> uint16_t, promoted to 'int' >> (nb_txq * nb_fwd_ports) -> result 'int' >> nb_fwd_lcores -> 'uint32_t' >> >> comparison between 'int' & 'uint32_t' gives warning. After some compiler >> version it is smart enough to not give a warning, but casting is >> required for old compilers. >> >> And back to my comment above, casting one of the parameter to >> 'lcoreid_t' also works, as it forcing promotion to 'unsigned int' >> instead. But logically casting looks odd, so keeping casting result. > > Where is the integer promotion happening? > Raslan reported following compile error, this version of the patch has the cast, but merged version, v3, doesn't. ``` ../../root/dpdk/app/test-pmd/config.c: In function 'icmp_echo_config_setup': ../../root/dpdk/app/test-pmd/config.c:5159:30: error: comparison between signed and unsigned integer expressions [-Werror=sign-compare] if ((nb_txq * nb_fwd_ports) < nb_fwd_lcores) ^ ``` > Also would be better to use strtoul() instead of atoi() when parsing the args. > That gives better error handling and handles someone passing negative number correctly. ^ permalink raw reply [flat|nested] 17+ messages in thread
* Re: [PATCH v2] app/testpmd: fix lcore ID restriction 2024-06-14 9:01 ` Ferruh Yigit @ 2024-06-14 15:27 ` Stephen Hemminger 2024-06-14 15:39 ` Ferruh Yigit 2024-06-14 17:50 ` Morten Brørup 0 siblings, 2 replies; 17+ messages in thread From: Stephen Hemminger @ 2024-06-14 15:27 UTC (permalink / raw) To: Ferruh Yigit Cc: Sivaprasad Tummala, david.marchand, aman.deep.singh, yuying.zhang, dev, stable On Fri, 14 Jun 2024 10:01:02 +0100 Ferruh Yigit <ferruh.yigit@amd.com> wrote: > On 6/13/2024 8:13 PM, Stephen Hemminger wrote: > > On Thu, 13 Jun 2024 17:51:14 +0100 > > Ferruh Yigit <ferruh.yigit@amd.com> wrote: > > > >>> Hi Sivaprasad, > >>> > >>> Is this '(lcoreid_t)' cast required? Because of integer promotion I > >>> think result will be correct without casting. > >>> > >>> (And without integer promotion considered, casting needs to be done on > >>> one of the variables, not to the result, because result may be already > >>> cast down I think. Anyway this is not required for this case since > >>> variables are u16.) > >>> > >> > >> Why casing required (for record) is, > >> 'nb_txq' -> uint16_t, promoted to 'int' > >> 'nb_fwd_ports' -> uint16_t, promoted to 'int' > >> (nb_txq * nb_fwd_ports) -> result 'int' > >> nb_fwd_lcores -> 'uint32_t' > >> > >> comparison between 'int' & 'uint32_t' gives warning. After some compiler > >> version it is smart enough to not give a warning, but casting is > >> required for old compilers. > >> > >> And back to my comment above, casting one of the parameter to > >> 'lcoreid_t' also works, as it forcing promotion to 'unsigned int' > >> instead. But logically casting looks odd, so keeping casting result. > > > > Where is the integer promotion happening? > > > > Raslan reported following compile error, this version of the patch has > the cast, but merged version, v3, doesn't. > > > ``` > ../../root/dpdk/app/test-pmd/config.c: In function 'icmp_echo_config_setup': > ../../root/dpdk/app/test-pmd/config.c:5159:30: error: comparison between > signed and unsigned integer expressions [-Werror=sign-compare] > if ((nb_txq * nb_fwd_ports) < nb_fwd_lcores) That does look like a compiler bug. uint16 multiplied by uint16 should be uint16. Not sure why DPDK keeps using small types like uint16 so much, doesn't have any real benefit since all this is in registers. ^ ^ permalink raw reply [flat|nested] 17+ messages in thread
* Re: [PATCH v2] app/testpmd: fix lcore ID restriction 2024-06-14 15:27 ` Stephen Hemminger @ 2024-06-14 15:39 ` Ferruh Yigit 2024-06-14 17:50 ` Morten Brørup 1 sibling, 0 replies; 17+ messages in thread From: Ferruh Yigit @ 2024-06-14 15:39 UTC (permalink / raw) To: Stephen Hemminger Cc: Sivaprasad Tummala, david.marchand, aman.deep.singh, yuying.zhang, dev, stable On 6/14/2024 4:27 PM, Stephen Hemminger wrote: > On Fri, 14 Jun 2024 10:01:02 +0100 > Ferruh Yigit <ferruh.yigit@amd.com> wrote: > >> On 6/13/2024 8:13 PM, Stephen Hemminger wrote: >>> On Thu, 13 Jun 2024 17:51:14 +0100 >>> Ferruh Yigit <ferruh.yigit@amd.com> wrote: >>> >>>>> Hi Sivaprasad, >>>>> >>>>> Is this '(lcoreid_t)' cast required? Because of integer promotion I >>>>> think result will be correct without casting. >>>>> >>>>> (And without integer promotion considered, casting needs to be done on >>>>> one of the variables, not to the result, because result may be already >>>>> cast down I think. Anyway this is not required for this case since >>>>> variables are u16.) >>>>> >>>> >>>> Why casing required (for record) is, >>>> 'nb_txq' -> uint16_t, promoted to 'int' >>>> 'nb_fwd_ports' -> uint16_t, promoted to 'int' >>>> (nb_txq * nb_fwd_ports) -> result 'int' >>>> nb_fwd_lcores -> 'uint32_t' >>>> >>>> comparison between 'int' & 'uint32_t' gives warning. After some compiler >>>> version it is smart enough to not give a warning, but casting is >>>> required for old compilers. >>>> >>>> And back to my comment above, casting one of the parameter to >>>> 'lcoreid_t' also works, as it forcing promotion to 'unsigned int' >>>> instead. But logically casting looks odd, so keeping casting result. >>> >>> Where is the integer promotion happening? >>> >> >> Raslan reported following compile error, this version of the patch has >> the cast, but merged version, v3, doesn't. >> >> >> ``` >> ../../root/dpdk/app/test-pmd/config.c: In function 'icmp_echo_config_setup': >> ../../root/dpdk/app/test-pmd/config.c:5159:30: error: comparison between >> signed and unsigned integer expressions [-Werror=sign-compare] >> if ((nb_txq * nb_fwd_ports) < nb_fwd_lcores) > > That does look like a compiler bug. uint16 multiplied by uint16 should be uint16. > Probably, this warning goes away with newer version of compiler. > Not sure why DPDK keeps using small types like uint16 so much, doesn't have any > real benefit since all this is in registers. ^ ^ permalink raw reply [flat|nested] 17+ messages in thread
* RE: [PATCH v2] app/testpmd: fix lcore ID restriction 2024-06-14 15:27 ` Stephen Hemminger 2024-06-14 15:39 ` Ferruh Yigit @ 2024-06-14 17:50 ` Morten Brørup 2024-06-14 18:11 ` Stephen Hemminger 1 sibling, 1 reply; 17+ messages in thread From: Morten Brørup @ 2024-06-14 17:50 UTC (permalink / raw) To: Stephen Hemminger, Ferruh Yigit Cc: Sivaprasad Tummala, david.marchand, aman.deep.singh, yuying.zhang, dev, stable > From: Stephen Hemminger [mailto:stephen@networkplumber.org] > Sent: Friday, 14 June 2024 17.27 > > On Fri, 14 Jun 2024 10:01:02 +0100 > Ferruh Yigit <ferruh.yigit@amd.com> wrote: > > > On 6/13/2024 8:13 PM, Stephen Hemminger wrote: > > > On Thu, 13 Jun 2024 17:51:14 +0100 > > > Ferruh Yigit <ferruh.yigit@amd.com> wrote: > > > > > >>> Hi Sivaprasad, > > >>> > > >>> Is this '(lcoreid_t)' cast required? Because of integer promotion > I > > >>> think result will be correct without casting. > > >>> > > >>> (And without integer promotion considered, casting needs to be > done on > > >>> one of the variables, not to the result, because result may be > already > > >>> cast down I think. Anyway this is not required for this case since > > >>> variables are u16.) > > >>> > > >> > > >> Why casing required (for record) is, > > >> 'nb_txq' -> uint16_t, promoted to 'int' > > >> 'nb_fwd_ports' -> uint16_t, promoted to 'int' > > >> (nb_txq * nb_fwd_ports) -> result 'int' > > >> nb_fwd_lcores -> 'uint32_t' > > >> > > >> comparison between 'int' & 'uint32_t' gives warning. After some > compiler > > >> version it is smart enough to not give a warning, but casting is > > >> required for old compilers. > > >> > > >> And back to my comment above, casting one of the parameter to > > >> 'lcoreid_t' also works, as it forcing promotion to 'unsigned int' > > >> instead. But logically casting looks odd, so keeping casting > result. > > > > > > Where is the integer promotion happening? > > > > > > > Raslan reported following compile error, this version of the patch has > > the cast, but merged version, v3, doesn't. > > > > > > ``` > > ../../root/dpdk/app/test-pmd/config.c: In function > 'icmp_echo_config_setup': > > ../../root/dpdk/app/test-pmd/config.c:5159:30: error: comparison > between > > signed and unsigned integer expressions [-Werror=sign-compare] > > if ((nb_txq * nb_fwd_ports) < nb_fwd_lcores) > > That does look like a compiler bug. uint16 multiplied by uint16 should > be uint16. Not, C doesn't promote types like that. Arithmetic operations always promote smaller types to int or unsigned int. And since uint16 fits in int, uint16 is promoted to int (not unsigned int). > Not sure why DPDK keeps using small types like uint16 so much, doesn't > have any > real benefit since all this is in registers. ^ permalink raw reply [flat|nested] 17+ messages in thread
* Re: [PATCH v2] app/testpmd: fix lcore ID restriction 2024-06-14 17:50 ` Morten Brørup @ 2024-06-14 18:11 ` Stephen Hemminger 0 siblings, 0 replies; 17+ messages in thread From: Stephen Hemminger @ 2024-06-14 18:11 UTC (permalink / raw) To: Morten Brørup Cc: Ferruh Yigit, Sivaprasad Tummala, david.marchand, aman.deep.singh, yuying.zhang, dev, stable On Fri, 14 Jun 2024 19:50:29 +0200 Morten Brørup <mb@smartsharesystems.com> wrote: > > > ``` > > > ../../root/dpdk/app/test-pmd/config.c: In function > > 'icmp_echo_config_setup': > > > ../../root/dpdk/app/test-pmd/config.c:5159:30: error: comparison > > between > > > signed and unsigned integer expressions [-Werror=sign-compare] > > > if ((nb_txq * nb_fwd_ports) < nb_fwd_lcores) > > > > That does look like a compiler bug. uint16 multiplied by uint16 should > > be uint16. > > Not, C doesn't promote types like that. Arithmetic operations always promote smaller types to int or unsigned int. And since uint16 fits in int, uint16 is promoted to int (not unsigned int). You are right. Long winded explanation here: https://stackoverflow.com/questions/46073295/implicit-type-promotion-rules ^ permalink raw reply [flat|nested] 17+ messages in thread
* [PATCH v3] app/testpmd: fix lcore ID restriction 2024-04-16 9:55 ` [PATCH v2] " Sivaprasad Tummala 2024-04-19 8:24 ` Ferruh Yigit 2024-04-19 11:30 ` Ferruh Yigit @ 2024-06-06 11:27 ` Sivaprasad Tummala 2024-06-11 23:37 ` Ferruh Yigit 2 siblings, 1 reply; 17+ messages in thread From: Sivaprasad Tummala @ 2024-06-06 11:27 UTC (permalink / raw) To: ferruh.yigit, david.marchand, aman.deep.singh; +Cc: dev, stable With modern CPUs, it is possible to have higher CPU count thus we can have higher RTE_MAX_LCORES. In testpmd application, the current config forwarding cores option "--nb-cores" is hard limited to 255. The patch fixes this constraint and also adjusts the lcore data structure to 32-bit to align with rte lcore APIs. Fixes: af75078fece3 ("first public release") Cc: stable@dpdk.org Signed-off-by: Sivaprasad Tummala <sivaprasad.tummala@amd.com> --- v3: * Removed unnecessary cast v2: * Fixed type mistmatch in comparison * Fixed incorrect format specifier --- app/test-pmd/config.c | 4 ++-- app/test-pmd/parameters.c | 4 ++-- app/test-pmd/testpmd.h | 2 +- 3 files changed, 5 insertions(+), 5 deletions(-) diff --git a/app/test-pmd/config.c b/app/test-pmd/config.c index ba1007ace6..a0c8bb8fe1 100644 --- a/app/test-pmd/config.c +++ b/app/test-pmd/config.c @@ -4785,9 +4785,9 @@ fwd_stream_on_other_lcores(uint16_t domain_id, lcoreid_t src_lc, continue; printf("Shared Rx queue group %u queue %hu can't be scheduled on different cores:\n", share_group, share_rxq); - printf(" lcore %hhu Port %hu queue %hu\n", + printf(" lcore %u Port %hu queue %hu\n", src_lc, src_port, src_rxq); - printf(" lcore %hhu Port %hu queue %hu\n", + printf(" lcore %u Port %hu queue %hu\n", lc_id, fs->rx_port, fs->rx_queue); printf("Please use --nb-cores=%hu to limit number of forwarding cores\n", nb_rxq); diff --git a/app/test-pmd/parameters.c b/app/test-pmd/parameters.c index c13f7564bf..22364e09ab 100644 --- a/app/test-pmd/parameters.c +++ b/app/test-pmd/parameters.c @@ -1071,8 +1071,8 @@ launch_args_parse(int argc, char** argv) break; case TESTPMD_OPT_NB_CORES_NUM: n = atoi(optarg); - if (n > 0 && n <= nb_lcores) - nb_fwd_lcores = (uint8_t) n; + if (n > 0 && (lcoreid_t)n <= nb_lcores) + nb_fwd_lcores = (lcoreid_t) n; else rte_exit(EXIT_FAILURE, "nb-cores should be > 0 and <= %d\n", diff --git a/app/test-pmd/testpmd.h b/app/test-pmd/testpmd.h index 0afae7d771..9facd7f281 100644 --- a/app/test-pmd/testpmd.h +++ b/app/test-pmd/testpmd.h @@ -84,7 +84,7 @@ extern volatile uint8_t f_quit; /* Maximum number of pools supported per Rx queue */ #define MAX_MEMPOOL 8 -typedef uint8_t lcoreid_t; +typedef uint32_t lcoreid_t; typedef uint16_t portid_t; typedef uint16_t queueid_t; typedef uint16_t streamid_t; -- 2.34.1 ^ permalink raw reply [flat|nested] 17+ messages in thread
* Re: [PATCH v3] app/testpmd: fix lcore ID restriction 2024-06-06 11:27 ` [PATCH v3] " Sivaprasad Tummala @ 2024-06-11 23:37 ` Ferruh Yigit 2024-06-13 16:36 ` Ferruh Yigit 0 siblings, 1 reply; 17+ messages in thread From: Ferruh Yigit @ 2024-06-11 23:37 UTC (permalink / raw) To: Sivaprasad Tummala, david.marchand, aman.deep.singh; +Cc: dev, stable On 6/6/2024 12:27 PM, Sivaprasad Tummala wrote: > With modern CPUs, it is possible to have higher > CPU count thus we can have higher RTE_MAX_LCORES. > In testpmd application, the current config forwarding > cores option "--nb-cores" is hard limited to 255. > > The patch fixes this constraint and also adjusts the lcore > data structure to 32-bit to align with rte lcore APIs. > > Fixes: af75078fece3 ("first public release") > Cc: stable@dpdk.org > > Signed-off-by: Sivaprasad Tummala <sivaprasad.tummala@amd.com> > Acked-by: Ferruh Yigit <ferruh.yigit@amd.com> Applied to dpdk-next-net/main, thanks. ^ permalink raw reply [flat|nested] 17+ messages in thread
* Re: [PATCH v3] app/testpmd: fix lcore ID restriction 2024-06-11 23:37 ` Ferruh Yigit @ 2024-06-13 16:36 ` Ferruh Yigit 0 siblings, 0 replies; 17+ messages in thread From: Ferruh Yigit @ 2024-06-13 16:36 UTC (permalink / raw) To: Sivaprasad Tummala, david.marchand, aman.deep.singh Cc: dev, stable, Raslan Darawsheh On 6/12/2024 12:37 AM, Ferruh Yigit wrote: > On 6/6/2024 12:27 PM, Sivaprasad Tummala wrote: >> With modern CPUs, it is possible to have higher >> CPU count thus we can have higher RTE_MAX_LCORES. >> In testpmd application, the current config forwarding >> cores option "--nb-cores" is hard limited to 255. >> >> The patch fixes this constraint and also adjusts the lcore >> data structure to 32-bit to align with rte lcore APIs. >> >> Fixes: af75078fece3 ("first public release") >> Cc: stable@dpdk.org >> >> Signed-off-by: Sivaprasad Tummala <sivaprasad.tummala@amd.com> >> > > Acked-by: Ferruh Yigit <ferruh.yigit@amd.com> > > Applied to dpdk-next-net/main, thanks. > Raslan reported build error because of the casting I asked to remove :( I will add it back in the next-net repo directly, meanwhile I will comment on the patch for details. ^ permalink raw reply [flat|nested] 17+ messages in thread
end of thread, other threads:[~2024-06-14 18:11 UTC | newest] Thread overview: 17+ messages (download: mbox.gz / follow: Atom feed) -- links below jump to the message on this page -- 2024-04-15 19:46 [PATCH] app/testpmd: fix lcore ID restriction Sivaprasad Tummala 2024-04-15 21:54 ` Stephen Hemminger 2024-04-15 21:56 ` Stephen Hemminger 2024-04-16 9:55 ` [PATCH v2] " Sivaprasad Tummala 2024-04-19 8:24 ` Ferruh Yigit 2024-04-19 11:30 ` Ferruh Yigit 2024-06-06 9:56 ` Tummala, Sivaprasad 2024-06-13 16:51 ` Ferruh Yigit 2024-06-13 19:13 ` Stephen Hemminger 2024-06-14 9:01 ` Ferruh Yigit 2024-06-14 15:27 ` Stephen Hemminger 2024-06-14 15:39 ` Ferruh Yigit 2024-06-14 17:50 ` Morten Brørup 2024-06-14 18:11 ` Stephen Hemminger 2024-06-06 11:27 ` [PATCH v3] " Sivaprasad Tummala 2024-06-11 23:37 ` Ferruh Yigit 2024-06-13 16:36 ` Ferruh Yigit
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).