s390x: smp: use CPU indexes instead of addresses
v2 cover s390x: smp: use CPU indexes instead of addresses
On s390x there are no guarantees about the CPU addresses, except that they shall be unique. This means that in some environments, it is theoretically possible that there is no match between the CPU address and its position (index) in the list of available CPUs returned by the system. Moreover, there are no guarantees about the ordering of the list, or even that it is consistent each time it is returned.
This series fixes a small bug in the SMP initialization code, adds a guarantee that the boot CPU will always have index 0, changes the existing smp_* functions to take indexes instead of addresses, and introduces some functions to allow tests to use CPU indexes instead of using hardcoded CPU addresses. This will allow the tests to run successfully in more environments (e.g. z/VM, LPAR).
Some existing tests are adapted to take advantage of the new functionalities.
v1->v2
* refactored the smp_* functions to accept indexes instead of addresses
* also fixed uv-host test
Claudio Imbrenda (6):
lib: s390x: smp: guarantee that boot CPU has index 0
lib: s390x: smp: refactor smp functions to accept indexes
s390x: smp: use CPU indexes instead of addresses
s390x: firq: use CPU indexes instead of addresses
s390x: skrf: use CPU indexes instead of addresses
s390x: uv-host: use CPU indexes instead of addresses
lib/s390x/smp.h | 20 +++--- lib/s390x/smp.c | 173 +++++++++++++++++++++++++++++------------------- s390x/firq.c | 26 ++------ s390x/skrf.c | 2 +- s390x/smp.c | 22 +++--- s390x/uv-host.c | 4 +- 6 files changed, 137 insertions(+), 110 deletions(-)
Signature · 1 line
2.34.1
PATCH v2 1/6 lib: s390x: smp: guarantee that boot CPU has index 0
Guarantee that the boot CPU has index 0. This simplifies the implementation of tests that require multiple CPUs.
Also fix a small bug in the allocation of the cpus array.
--- lib/s390x/smp.c | 27 +++++++++++++++++++-------- 1 file changed, 19 insertions(+), 8 deletions(-)
lib/s390x/smp.c +19 -8
@@ -25,7 +25,6 @@#include "sclp.h" static struct cpu *cpus; -static struct cpu *cpu0;static struct spinlock lock;extern void smp_cpu_setup_state (void );
@@ -69,7 +68,7 @@ static int smp_cpu_stop_nolock(uint16_t addr, bool store)uint8_t order = store ?SIGP_STOP_AND_STORE_STATUS :SIGP_STOP ; cpu =smp_cpu_from_addr (addr); -if (!cpu || cpu == cpu0) +if (!cpu || addr == cpus[0 ].addr)return -1 ;if (sigp_retry (addr, order,0 ,NULL ))
@@ -193,7 +192,7 @@ int smp_cpu_setup(uint16_t addr, struct psw psw)sigp_retry (cpu->addr,SIGP_SET_PREFIX , (unsigned long )lc,NULL );/* Copy all exception psws. */ -memcpy (lc, cpu0->lowcore,512 ); +memcpy (lc, cpus[0 ].lowcore,512 );/* Setup stack */ cpu->stack = (uint64_t *)alloc_pages (2 );
@@ -251,15 +250,27 @@ void smp_setup(void)if (num >1 )printf ("SMP: Initializing, found %d cpus\n" , num); - cpus =calloc (num,sizeof (cpus)); + cpus =calloc (num,sizeof (*cpus));for (i =0 ; i < num; i++) { cpus[i].addr = entry[i].address; cpus[i].active = false; +/* +* Fill in the boot CPU. If the boot CPU is not at index 0, +* swap it with the one at index 0. This guarantees that the +* boot CPU will always have index 0. If the boot CPU was +* already at index 0, a few extra useless assignments are +* performed, but everything will work ok. +* Notice that there is no guarantee that the list of CPUs +* returned by the Read SCP Info command is in any +* particular order, or that its order will stay consistent +* across multiple invocations. +*/ if (entry[i].address == cpu0_addr) { - cpu0 = &cpus[i]; - cpu0->stack = stackptr; - cpu0->lowcore = (void *)0 ; - cpu0->active = true; + cpus[i].addr = cpus[0 ].addr; + cpus[0 ].addr = cpu0_addr; + cpus[0 ].stack = stackptr; + cpus[0 ].lowcore = (void *)0 ; + cpus[0 ].active = true; } }spin_unlock (&lock);
Signature · 1 line
2.34.1
Re: [kvm-unit-tests PATCH v2 1/6] lib: s390x: smp: guarantee that boot CPU has index 0
Quoted 6 lines · Claudio Imbrenda
Guarantee that the boot CPU has index 0. This simplifies the implementation of tests that require multiple CPUs.
Also fix a small bug in the allocation of the cpus array.
Signed-off-by Claudio Imbrenda <[email protected]>
Re: [kvm-unit-tests PATCH v2 1/6] lib: s390x: smp: guarantee that boot CPU has index 0
In reply to Claudio ImbrendaQuoted 6 lines · Claudio Imbrenda
Guarantee that the boot CPU has index 0. This simplifies the implementation of tests that require multiple CPUs.
Also fix a small bug in the allocation of the cpus array.
Signed-off-by Claudio Imbrenda <[email protected]>
PATCH v2 2/6 lib: s390x: smp: refactor smp functions to accept indexes
In reply to Claudio ImbrendaRefactor all the smp_* functions to accept CPU indexes instead of CPU addresses.
Add SIGP wrappers to use indexes instead of addresses. Raw SIGP calls using addresses are still possible.
Add a few other useful functions to deal with CPU indexes.
--- lib/s390x/smp.h | 20 ++++--- lib/s390x/smp.c | 148 ++++++++++++++++++++++++++++-------------------- 2 files changed, 99 insertions(+), 69 deletions(-)
lib/s390x/smp.h +12 -8
@@ -37,15 +37,19 @@ struct cpu_status {int smp_query_num_cpus (void );struct cpu *smp_cpu_from_addr (uint16_t addr); -bool smp_cpu_stopped (uint16_t addr); -bool smp_sense_running_status (uint16_t addr); -int smp_cpu_restart (uint16_t addr); -int smp_cpu_start (uint16_t addr,struct psw psw); -int smp_cpu_stop (uint16_t addr); -int smp_cpu_stop_store_status (uint16_t addr); -int smp_cpu_destroy (uint16_t addr); -int smp_cpu_setup (uint16_t addr,struct psw psw); +struct cpu *smp_cpu_from_idx (uint16_t idx); +uint16_t smp_cpu_addr (uint16_t idx); +bool smp_cpu_stopped (uint16_t idx); +bool smp_sense_running_status (uint16_t idx); +int smp_cpu_restart (uint16_t idx); +int smp_cpu_start (uint16_t idx,struct psw psw); +int smp_cpu_stop (uint16_t idx); +int smp_cpu_stop_store_status (uint16_t idx); +int smp_cpu_destroy (uint16_t idx); +int smp_cpu_setup (uint16_t idx,struct psw psw);void smp_teardown (void );void smp_setup (void ); +int smp_sigp (uint16_t idx,uint8_t order,unsigned long parm,uint32_t *status); +int smp_sigp_retry (uint16_t idx,uint8_t order,unsigned long parm,uint32_t *status);#endif
lib/s390x/smp.c +87 -61
272 lines not on this page · raw message
Signature · 1 line
2.34.1
Re: [kvm-unit-tests PATCH v2 2/6] lib: s390x: smp: refactor smp functions to accept indexes
Quoted 9 lines · Claudio Imbrenda
Refactor all the smp_* functions to accept CPU indexes instead of CPU addresses.
Add SIGP wrappers to use indexes instead of addresses. Raw SIGP calls using addresses are still possible.
Add a few other useful functions to deal with CPU indexes.
Signed-off-by Claudio Imbrenda <[email protected]>
Re: [kvm-unit-tests PATCH v2 2/6] lib: s390x: smp: refactor smp functions to accept indexes
In reply to Claudio ImbrendaQuoted 178 lines · Claudio Imbrenda
Refactor all the smp_* functions to accept CPU indexes instead of CPU addresses.
Add SIGP wrappers to use indexes instead of addresses. Raw SIGP calls using addresses are still possible.
Add a few other useful functions to deal with CPU indexes.
Signed-off-by Claudio Imbrenda <[email protected]>--- lib/s390x/smp.h | 20 ++++--- lib/s390x/smp.c | 148 ++++++++++++++++++++++++++++-------------------- 2 files changed, 99 insertions(+), 69 deletions(-)lib/s390x/smp.h +12
-8@@ -37,15 +37,19 @@ struct cpu_status {int smp_query_num_cpus (void );struct cpu *smp_cpu_from_addr (uint16_t addr); -bool smp_cpu_stopped (uint16_t addr); -bool smp_sense_running_status (uint16_t addr); -int smp_cpu_restart (uint16_t addr); -int smp_cpu_start (uint16_t addr,struct psw psw); -int smp_cpu_stop (uint16_t addr); -int smp_cpu_stop_store_status (uint16_t addr); -int smp_cpu_destroy (uint16_t addr); -int smp_cpu_setup (uint16_t addr,struct psw psw); +struct cpu *smp_cpu_from_idx (uint16_t idx); +uint16_t smp_cpu_addr (uint16_t idx); +bool smp_cpu_stopped (uint16_t idx); +bool smp_sense_running_status (uint16_t idx); +int smp_cpu_restart (uint16_t idx); +int smp_cpu_start (uint16_t idx,struct psw psw); +int smp_cpu_stop (uint16_t idx); +int smp_cpu_stop_store_status (uint16_t idx); +int smp_cpu_destroy (uint16_t idx); +int smp_cpu_setup (uint16_t idx,struct psw psw);void smp_teardown (void );void smp_setup (void ); +int smp_sigp (uint16_t idx,uint8_t order,unsigned long parm,uint32_t *status); +int smp_sigp_retry (uint16_t idx,uint8_t order,unsigned long parm,uint32_t *status);#endif lib/s390x/smp.c +48
-21@@ -29,11 +29,28 @@ static struct spinlock lock;extern void smp_cpu_setup_state (void ); +static void check_idx (uint16_t idx) +{ +assert (idx <smp_query_num_cpus ()); +} +int smp_query_num_cpus (void ) {return sclp_get_cpu_num (); } +int smp_sigp (uint16_t idx,uint8_t order,unsigned long parm,uint32_t *status) +{ +check_idx (idx); +return sigp (cpus[idx].addr, order, parm, status); +} + +int smp_sigp_retry (uint16_t idx,uint8_t order,unsigned long parm,uint32_t *status) +{ +check_idx (idx); +return sigp_retry (cpus[idx].addr, order, parm, status); +} +struct cpu *smp_cpu_from_addr (uint16_t addr) {int i, num =smp_query_num_cpus ();@@ -45,174 +62,183 @@ struct cpu *smp_cpu_from_addr(uint16_t addr)return NULL ; } > -bool smp_cpu_stopped (uint16_t addr) +struct cpu *smp_cpu_from_idx (uint16_t idx) +{ +check_idx (idx); +return &cpus[idx]; +} + +uint16_t smp_cpu_addr (uint16_t idx) +{ +check_idx (idx); +return cpus[idx].addr; +} + +bool smp_cpu_stopped (uint16_t idx) {uint32_t status; -if (sigp (addr,SIGP_SENSE ,0 , &status) !=SIGP_CC_STATUS_STORED ) +if (smp_sigp (idx,SIGP_SENSE ,0 , &status) !=SIGP_CC_STATUS_STORED )return false;return !!(status & (SIGP_STATUS_CHECK_STOP |SIGP_STATUS_STOPPED )); } -bool smp_sense_running_status (uint16_t addr) +bool smp_sense_running_status (uint16_t idx) { -if (sigp (addr,SIGP_SENSE_RUNNING ,0 ,NULL ) !=SIGP_CC_STATUS_STORED ) +if (smp_sigp (idx,SIGP_SENSE_RUNNING ,0 ,NULL ) !=SIGP_CC_STATUS_STORED )return true;/* Status stored condition code is equivalent to cpu not running. */ return false; } -static int smp_cpu_stop_nolock (uint16_t addr,bool store) +static int smp_cpu_stop_nolock (uint16_t idx,bool store) { -struct cpu *cpu;uint8_t order = store ?SIGP_STOP_AND_STORE_STATUS :SIGP_STOP ; - cpu =smp_cpu_from_addr (addr); -if (!cpu || addr == cpus[0 ].addr) +/* refuse to work on the boot CPU */ +if (idx ==0 )return -1 ; -if (sigp_retry (addr, order,0 ,NULL )) +if (smp_sigp_retry (idx, order,0 ,NULL ))return -1 ; -while (!smp_cpu_stopped (addr)) +while (!smp_cpu_stopped (idx))mb (); - cpu->active = false; +/* idx has been already checked by the smp_* functions called above */ + cpus[idx].active = false;return 0 ; } -int smp_cpu_stop (uint16_t addr) +int smp_cpu_stop (uint16_t idx) {int rc;spin_lock (&lock); - rc =smp_cpu_stop_nolock (addr, false); + rc =smp_cpu_stop_nolock (idx, false);spin_unlock (&lock);return rc; } -int smp_cpu_stop_store_status (uint16_t addr) +int smp_cpu_stop_store_status (uint16_t idx) {int rc;spin_lock (&lock); - rc =smp_cpu_stop_nolock (addr, true); + rc =smp_cpu_stop_nolock (idx, true);spin_unlock (&lock);return rc; } -static int smp_cpu_restart_nolock (uint16_t addr,struct psw *psw) +static int smp_cpu_restart_nolock (uint16_t idx,struct psw *psw) {int rc; -struct cpu *cpu =smp_cpu_from_addr (addr); -if (!cpu) -return -1 ; +check_idx (idx);if (psw) { - cpu->lowcore->restart_new_psw.mask = psw->mask; - cpu->lowcore->restart_new_psw.addr = psw->addr; + cpus[idx].lowcore->restart_new_psw.mask = psw->mask; + cpus[idx].lowcore->restart_new_psw.addr = psw->addr;
lib/s390x/smp.c struct cpu *smp_cpu_from_addr(uint16_t addr) · line 146
Quoted 10 lines
lib/s390x/smp.c +2
-2}/* * Stop the cpu, so we don't have a race between a running cpu * and the restart in the test that checks if the cpu is * running after the restart. */ -smp_cpu_stop_nolock (addr, false); - rc =sigp (addr,SIGP_RESTART ,0 ,NULL ); +smp_cpu_stop_nolock (idx, false); + rc =sigp (cpus[idx].addr,SIGP_RESTART ,0 ,NULL );
What about using the smp wrapper 'smp_sigp(idx, SIGP_RESTART, 0, NULL)' here as well?
Quoted 73 lines
lib/s390x/smp.c +13
-22if (rc)return rc;/* * The order has been accepted, but the actual restart may not * have been performed yet, so wait until the cpu is running. */ -while (smp_cpu_stopped (addr)) +while (smp_cpu_stopped (idx))mb (); - cpu->active = true; + cpus[idx].active = true;return 0 ; } -int smp_cpu_restart (uint16_t addr) +int smp_cpu_restart (uint16_t idx) {int rc;spin_lock (&lock); - rc =smp_cpu_restart_nolock (addr,NULL ); + rc =smp_cpu_restart_nolock (idx,NULL );spin_unlock (&lock);return rc; } -int smp_cpu_start (uint16_t addr,struct psw psw) +int smp_cpu_start (uint16_t idx,struct psw psw) {int rc;spin_lock (&lock); - rc =smp_cpu_restart_nolock (addr, &psw); + rc =smp_cpu_restart_nolock (idx, &psw);spin_unlock (&lock);return rc; } -int smp_cpu_destroy (uint16_t addr) +int smp_cpu_destroy (uint16_t idx) { -struct cpu *cpu;int rc;spin_lock (&lock); - rc =smp_cpu_stop_nolock (addr, false); + rc =smp_cpu_stop_nolock (idx, false);if (!rc) { - cpu =smp_cpu_from_addr (addr); -free_pages (cpu->lowcore); -free_pages (cpu->stack); - cpu->lowcore = (void *)-1UL ; - cpu->stack = (void *)-1UL ; +free_pages (cpus[idx].lowcore); +free_pages (cpus[idx].stack); + cpus[idx].lowcore = (void *)-1UL ; + cpus[idx].stack = (void *)-1UL ; }spin_unlock (&lock);return rc; } -int smp_cpu_setup (uint16_t addr,struct psw psw) +static int smp_cpu_setup_nolock (uint16_t idx,struct psw psw) {struct lowcore *lc; -struct cpu *cpu; -int rc =-1 ; - -spin_lock (&lock); - -if (!cpus) - goto out;
lib/s390x/smp.c struct cpu *smp_cpu_from_addr(uint16_t addr) · line 203
Quoted 10 lines
lib/s390x/smp.c +3
-5- cpu =smp_cpu_from_addr (addr); - -if (!cpu || cpu->active) - goto out; +if (cpus[idx].active) +return -1 ; -sigp_retry (cpu->addr,SIGP_INITIAL_CPU_RESET ,0 ,NULL ); +sigp_retry (cpus[idx].addr,SIGP_INITIAL_CPU_RESET ,0 ,NULL );
You may want to use the smp wrapper 'smp_sigp_retry' here.
lib/s390x/smp.c struct cpu *smp_cpu_from_addr(uint16_t addr) · line 207
Quoted 7 lines
lib/s390x/smp.c +2
-3lc =alloc_pages_flags (1 ,AREA_DMA31 ); - cpu->lowcore = lc; -memset (lc,0 ,PAGE_SIZE *2 ); -sigp_retry (cpu->addr,SIGP_SET_PREFIX , (unsigned long )lc,NULL ); + cpus[idx].lowcore = lc; +sigp_retry (cpus[idx].addr,SIGP_SET_PREFIX , (unsigned long )lc,NULL );
smp_sigp_retry
Quoted 36 lines
lib/s390x/smp.c +13
-8/* Copy all exception psws. */ memcpy (lc, cpus[0 ].lowcore,512 );/* Setup stack */ - cpu->stack = (uint64_t *)alloc_pages (2 ); + cpus[idx].stack = (uint64_t *)alloc_pages (2 );/* Start without DAT and any other mask bits. */ - cpu->lowcore->sw_int_psw.mask = psw.mask; - cpu->lowcore->sw_int_psw.addr = psw.addr; - cpu->lowcore->sw_int_grs[14 ] = psw.addr; - cpu->lowcore->sw_int_grs[15 ] = (uint64_t )cpu->stack + (PAGE_SIZE *4 ); + lc->sw_int_psw.mask = psw.mask; + lc->sw_int_psw.addr = psw.addr; + lc->sw_int_grs[14 ] = psw.addr; + lc->sw_int_grs[15 ] = (uint64_t )cpus[idx].stack + (PAGE_SIZE *4 ); lc->restart_new_psw.mask =PSW_MASK_64 ; lc->restart_new_psw.addr = (uint64_t )smp_cpu_setup_state; lc->sw_int_crs[0 ] =BIT_ULL (CTL0_AFP );/* Start processing */ -smp_cpu_restart_nolock (addr,NULL ); +smp_cpu_restart_nolock (idx,NULL );/* Wait until the cpu has finished setup and started the provided psw */ while (lc->restart_new_psw.addr != psw.addr)mb (); - rc =0 ; -out: + +return 0 ; +} + +int smp_cpu_setup (uint16_t idx,struct psw psw) +{ +int rc =-1 ;
lib/s390x/smp.c struct cpu *smp_cpu_from_addr(uint16_t addr)
Quoted 10 lines
lib/s390x/smp.c +6
-0+ +spin_lock (&lock); +if (cpus) { +check_idx (idx); + rc =smp_cpu_setup_nolock (idx, psw); + }spin_unlock (&lock);return rc; }
With my nits fixed:
Re: [kvm-unit-tests PATCH v2 2/6] lib: s390x: smp: refactor smp functions to accept indexes
On Tue, 15 Feb 2022 12:09:53 +0100
Steffen Eiden <[email protected]> wrote:
[...]
Quoted 2 lines
What about using the smp wrapper '
smp_sigp(idx, SIGP_RESTART, 0, NULL)' here as well?
[...]
Quoted 1 line
With my nits fixed:
maybe I should add a comment explaining why I did not use the smp_ variants.
the reason is that the smp_ variants check the validity of the CPU index. but in those places, we have already checked (directly or indirectly) that the index is valid, so I save one useless check.
on the other hand, I don't know if it makes sense to optimize for that, since it's not a hot path, and one extra check will not kill the performance.
Quoted 2 lines
Reviewed-by Steffen Eiden <[email protected]>
Re: [kvm-unit-tests PATCH v2 2/6] lib: s390x: smp: refactor smp functions to accept indexes
Quoted 15 lines · Claudio Imbrenda
On Tue, 15 Feb 2022 12:09:53 +0100
Steffen Eiden <[email protected]> wrote:[...]
Quoted 2 lines
What about using the smp wrapper '
smp_sigp(idx, SIGP_RESTART, 0, NULL)' here as well?[...]
Quoted 1 line
With my nits fixed:
maybe I should add a comment explaining why I did not use the
smp_variants.
Quoted 7 lines
the reason is that the
smp_variants check the validity of the CPU index. but in those places, we have already checked (directly or indirectly) that the index is valid, so I save one useless check.Quoted 1 line
on the other hand, I don't know if it makes sense to optimize for that,
since it's not a hot path, and one extra check will not kill the performance.
I would prefer the use of the smp_ variant. The extra assert won't clutter the output and the code is more consistent. However, a short comment is also fine for me if you prefer that.
Quoted 3 lines
Quoted 2 lines
Reviewed-by Steffen Eiden <[email protected]>
Re: [kvm-unit-tests PATCH v2 2/6] lib: s390x: smp: refactor smp functions to accept indexes
Quoted 19 lines · On Tue, 15 Feb 2022 12:43:15 +0100 Steffen Eiden <[email protected]> wrote:
Quoted 18 lines · Claudio Imbrenda
On Tue, 15 Feb 2022 12:09:53 +0100
Steffen Eiden <[email protected]> wrote:[...]
Quoted 2 lines
What about using the smp wrapper '
smp_sigp(idx, SIGP_RESTART, 0, NULL)' here as well?[...]
Quoted 1 line
With my nits fixed:
maybe I should add a comment explaining why I did not use the
smp_variants.the reason is that the
smp_variants check the validity of the CPU index. but in those places, we have already checked (directly or indirectly) that the index is valid, so I save one useless check.
Quoted 7 lines
Quoted 4 lines
Quoted 1 line
on the other hand, I don't know if it makes sense to optimize for that,
since it's not a hot path, and one extra check will not kill the performance.
I would prefer the use of the
smp_variant. The extra assert won't clutter the output and the code is more consistent. However, a short comment is also fine for me if you prefer that.
I guess I'll use the smp_ variant and add a few lines in the patch description to explain that we're doing some extra checks, but the code is more readable
Quoted 4 lines
Quoted 4 lines
Quoted 3 lines
Reviewed-by Steffen Eiden <[email protected]>
Re: [kvm-unit-tests PATCH v2 2/6] lib: s390x: smp: refactor smp functions to accept indexes
Quoted 26 lines · Claudio Imbrenda
Quoted 23 lines · On Tue, 15 Feb 2022 12:43:15 +0100 Steffen Eiden <[email protected]> wrote:
Quoted 22 lines · Claudio Imbrenda
On Tue, 15 Feb 2022 12:09:53 +0100
Steffen Eiden <[email protected]> wrote:[...]
Quoted 2 lines
What about using the smp wrapper '
smp_sigp(idx, SIGP_RESTART, 0, NULL)' here as well?[...]
Quoted 1 line
With my nits fixed:
maybe I should add a comment explaining why I did not use the
smp_variants.the reason is that the
smp_variants check the validity of the CPU index. but in those places, we have already checked (directly or indirectly) that the index is valid, so I save one useless check.Quoted 1 line
on the other hand, I don't know if it makes sense to optimize for that,
since it's not a hot path, and one extra check will not kill the performance.
Quoted 13 lines
Quoted 3 lines
I would prefer the use of the
smp_variant. The extra assert won't clutter the output and the code is more consistent. However, a short comment is also fine for me if you prefer that.I guess I'll use the
smp_variant and add a few lines in the patch description to explain that we're doing some extra checks, but the code is more readableQuoted 4 lines
Quoted 4 lines
Quoted 3 lines
Reviewed-by Steffen Eiden <[email protected]>
Doesn't make a difference to me as you use cpu.addr in the sigp_ which tells me it's a cpu address and not an idx.
PATCH v2 3/6 s390x: smp: use CPU indexes instead of addresses
In reply to Claudio ImbrendaAdapt the test to the new semantics of the smp_* functions, and use CPU indexes instead of addresses.
--- s390x/smp.c | 22 +++++++++++----------- 1 file changed, 11 insertions(+), 11 deletions(-)
s390x/smp.c +11 -11
@@ -56,7 +56,7 @@ static void test_start(void)*/ static void test_restart (void ) { -struct cpu *cpu =smp_cpu_from_addr (1 ); +struct cpu *cpu =smp_cpu_from_idx (1 );struct lowcore *lc = cpu->lowcore; lc->restart_new_psw.mask =extract_psw_mask ();
@@ -92,7 +92,7 @@ static void test_stop(void)static void test_stop_store_status (void ) { -struct cpu *cpu =smp_cpu_from_addr (1 ); +struct cpu *cpu =smp_cpu_from_idx (1 );struct lowcore *lc = (void *)0x0 ;report_prefix_push ("stop store status" );
@@ -129,7 +129,7 @@ static void test_store_status(void)report_prefix_push ("running" );smp_cpu_restart (1 ); -sigp (1 ,SIGP_STORE_STATUS_AT_ADDRESS , (uintptr_t )status, &r); +smp_sigp (1 ,SIGP_STORE_STATUS_AT_ADDRESS , (uintptr_t )status, &r);report (r ==SIGP_STATUS_INCORRECT_STATE ,"incorrect state" );report (!memcmp (status, (void *)status +PAGE_SIZE ,PAGE_SIZE ),"status not written" );
@@ -138,7 +138,7 @@ static void test_store_status(void)memset (status,0 ,PAGE_SIZE );report_prefix_push ("stopped" );smp_cpu_stop (1 ); -sigp (1 ,SIGP_STORE_STATUS_AT_ADDRESS , (uintptr_t )status,NULL ); +smp_sigp (1 ,SIGP_STORE_STATUS_AT_ADDRESS , (uintptr_t )status,NULL );while (!status->prefix) {mb (); }report_pass ("status written" );free_pages (status);
@@ -176,7 +176,7 @@ static void test_ecall(void)smp_cpu_start (1 , psw);wait_for_flag ();set_flag (0 ); -sigp (1 ,SIGP_EXTERNAL_CALL ,0 ,NULL ); +smp_sigp (1 ,SIGP_EXTERNAL_CALL ,0 ,NULL );wait_for_flag ();smp_cpu_stop (1 );report_prefix_pop ();
@@ -210,7 +210,7 @@ static void test_emcall(void)smp_cpu_start (1 , psw);wait_for_flag ();set_flag (0 ); -sigp (1 ,SIGP_EMERGENCY_SIGNAL ,0 ,NULL ); +smp_sigp (1 ,SIGP_EMERGENCY_SIGNAL ,0 ,NULL );wait_for_flag ();smp_cpu_stop (1 );report_prefix_pop ();
@@ -253,8 +253,8 @@ static void test_reset_initial(void)smp_cpu_start (1 , psw);wait_for_flag (); -sigp_retry (1 ,SIGP_INITIAL_CPU_RESET ,0 ,NULL ); -sigp (1 ,SIGP_STORE_STATUS_AT_ADDRESS , (uintptr_t )status,NULL ); +smp_sigp_retry (1 ,SIGP_INITIAL_CPU_RESET ,0 ,NULL ); +smp_sigp (1 ,SIGP_STORE_STATUS_AT_ADDRESS , (uintptr_t )status,NULL );report_prefix_push ("clear" );report (!status->psw.mask && !status->psw.addr,"psw" );
@@ -299,11 +299,11 @@ static void test_reset(void) psw.addr = (unsigned long )test_func;report_prefix_push ("cpu reset" ); -sigp (1 ,SIGP_EMERGENCY_SIGNAL ,0 ,NULL ); -sigp (1 ,SIGP_EXTERNAL_CALL ,0 ,NULL ); +smp_sigp (1 ,SIGP_EMERGENCY_SIGNAL ,0 ,NULL ); +smp_sigp (1 ,SIGP_EXTERNAL_CALL ,0 ,NULL );smp_cpu_start (1 , psw); -sigp_retry (1 ,SIGP_CPU_RESET ,0 ,NULL ); +smp_sigp_retry (1 ,SIGP_CPU_RESET ,0 ,NULL );report (smp_cpu_stopped (1 ),"cpu stopped" );set_flag (0 );
Signature · 1 line
2.34.1
Quoted 5 lines · Claudio Imbrenda
Adapt the test to the new semantics of the smp_* functions, and use CPU
indexes instead of addresses.Signed-off-by Claudio Imbrenda <[email protected]>
Quoted 4 lines · Claudio Imbrenda
Adapt the test to the new semantics of the smp_* functions, and use CPU indexes instead of addresses.
Signed-off-by Claudio Imbrenda <[email protected]>
PATCH v2 4/6 s390x: firq: use CPU indexes instead of addresses
In reply to Claudio ImbrendaAdapt the test to the new semantics of the smp_* functions, and use CPU indexes instead of addresses.
replace the checks with asserts, the 3 CPUs are guaranteed to be there.
--- s390x/firq.c | 26 ++++++-------------------- 1 file changed, 6 insertions(+), 20 deletions(-)
s390x/firq.c +6 -20
@@ -44,24 +44,13 @@ static void test_wait_state_delivery(void) goto out; } -if (stap ()) { -report_skip ("need to start on CPU #0" ); - goto out; - } - -/* -* We want CPU #2 to be stopped. This should be the case at this -* point, however, we want to sense if it even exists as well. -*/ +/* Stop CPU #2. It must succeed because we have at least 3 CPUs */ ret =smp_cpu_stop (2 ); -if (ret) { -report_skip ("CPU #2 not found" ); - goto out; - } +assert (!ret);/* -* We're going to perform an SCLP service call but expect -* the interrupt on CPU #1 while it is in the wait state. +* We're going to perform an SCLP service call but expect the +* interrupt on CPU #1 while it is in the wait state. */ sclp_mark_busy ();
@@ -69,11 +58,8 @@ static void test_wait_state_delivery(void) psw.mask =extract_psw_mask (); psw.addr = (unsigned long )wait_for_sclp_int; ret =smp_cpu_setup (1 , psw); -if (ret) { -sclp_clear_busy (); -report_skip ("cpu #1 not found" ); - goto out; - } +/* This must not fail because we have at least 3 CPUs */ +assert (!ret);/* * We'd have to jump trough some hoops to sense e.g., via SIGP
Signature · 1 line
2.34.1
Re: [kvm-unit-tests PATCH v2 4/6] s390x: firq: use CPU indexes instead of addresses
Quoted 8 lines · Claudio Imbrenda
Adapt the test to the new semantics of the smp_* functions, and use CPU
indexes instead of addresses.replace the checks with asserts, the 3 CPUs are guaranteed to be there.
Signed-off-by Claudio Imbrenda <[email protected]>
Re: [kvm-unit-tests PATCH v2 4/6] s390x: firq: use CPU indexes instead of addresses
In reply to Claudio ImbrendaQuoted 6 lines · Claudio Imbrenda
Adapt the test to the new semantics of the smp_* functions, and use CPU indexes instead of addresses.
replace the checks with asserts, the 3 CPUs are guaranteed to be there.
Signed-off-by Claudio Imbrenda <[email protected]>
PATCH v2 5/6 s390x: skrf: use CPU indexes instead of addresses
In reply to Claudio ImbrendaAdapt the test to the new semantics of the smp_* functions, and use CPU indexes instead of addresses.
--- s390x/skrf.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
s390x/skrf.c +1 -1
@@ -176,7 +176,7 @@ static void test_exception_ext_new(void)wait_for_flag ();set_flag (0 ); -sigp (1 ,SIGP_EXTERNAL_CALL ,0 ,NULL ); +smp_sigp (1 ,SIGP_EXTERNAL_CALL ,0 ,NULL );wait_for_flag ();smp_cpu_stop (1 );report_prefix_pop ();
Signature · 1 line
2.34.1
Re: [kvm-unit-tests PATCH v2 5/6] s390x: skrf: use CPU indexes instead of addresses
Quoted 5 lines · Claudio Imbrenda
Adapt the test to the new semantics of the smp_* functions, and use CPU
indexes instead of addresses.Signed-off-by Claudio Imbrenda <[email protected]>
Re: [kvm-unit-tests PATCH v2 5/6] s390x: skrf: use CPU indexes instead of addresses
In reply to Claudio ImbrendaQuoted 4 lines · Claudio Imbrenda
Adapt the test to the new semantics of the smp_* functions, and use CPU indexes instead of addresses.
Signed-off-by Claudio Imbrenda <[email protected]>
PATCH v2 6/6 s390x: uv-host: use CPU indexes instead of addresses
In reply to Claudio ImbrendaAdapt the test to the new semantics of the smp_* functions, and use CPU indexes instead of addresses.
--- s390x/uv-host.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)
s390x/uv-host.c +2 -2
@@ -267,12 +267,12 @@ static void test_config_create(void) uvcb_cgc.conf_base_stor_origin = tmp;if (smp_query_num_cpus () ==1 ) { -sigp_retry (1 ,SIGP_SET_PREFIX , +smp_sigp_retry (1 ,SIGP_SET_PREFIX , uvcb_cgc.conf_var_stor_origin +PAGE_SIZE ,NULL ); rc =uv_call (0 , (uint64_t )&uvcb_cgc);report (uvcb_cgc.header.rc ==0x10e && rc ==1 && !uvcb_cgc.guest_handle,"variable storage area contains lowcore" ); -sigp_retry (1 ,SIGP_SET_PREFIX ,0x0 ,NULL ); +smp_sigp_retry (1 ,SIGP_SET_PREFIX ,0x0 ,NULL ); } tmp = uvcb_cgc.guest_sca;
Signature · 1 line
2.34.1
Re: [kvm-unit-tests PATCH v2 6/6] s390x: uv-host: use CPU indexes instead of addresses
Quoted 15 lines · Claudio Imbrenda
Adapt the test to the new semantics of the smp_* functions, and use CPU
indexes instead of addresses.Signed-off-by Claudio Imbrenda <[email protected]>---
s390x/uv-host.c | 4 ++--1 file changed, 2 insertions(+), 2 deletions(-)s390x/uv-host.c +0
-0@@ -267,12 +267,12 @@ static void test_config_create(void) uvcb_cgc.conf_base_stor_origin = tmp;
Quoted 4 lines
if (smp_query_num_cpus()== 1) {
- sigp_retry(1,SIGP_SET_PREFIX, + smp_sigp_retry(1,SIGP_SET_PREFIX,
As smp_query_num_cpus() == 1, smp_sigp* will always run into an assert() here.
Re: [kvm-unit-tests PATCH v2 6/6] s390x: uv-host: use CPU indexes instead of addresses
In reply to Claudio ImbrendaQuoted 4 lines · Claudio Imbrenda
Adapt the test to the new semantics of the smp_* functions, and use CPU indexes instead of addresses.
Signed-off-by Claudio Imbrenda <[email protected]