arm/arm64: Add ITS tests
Subsystems (3)
v3 cover arm/arm64: Add ITS tests
This series is a revival of an RFC series sent in Dec 2016 [1]. Given the amount of code and the lack of traction at that time, I haven't respinned until now. However a recent bug found related to the ITS migration convinced me that this work may deserve to be respinned and enhanced.
Tests exercise main ITS commands and also test migration. With the migration framework, we are able to trigger the migration from guest and that is very practical actually.
What is particular with the ITS programming is that most of the commands are passed through queues and there is real error handling. Invalid commands are just ignored and that is not really tester friendly.
This series includes Andre's patch: "arm: gic: Provide per-IRQ helper functions" [2]
test_migrate_unmapped_collection is currently failing with upstream kernel. See [3].
The series can be fount at:
https://github.com/eauger/kut/tree/its-v3
Best Regards
Eric
History:
v2 -> v3:
- fix 32b compilation
- take into account Drew's comments (see individual diff logs)
v1 -> v2:
- took into account Zenghui's comments
- collect R-b's from Thomas
References:
[1] [kvm-unit-tests RFC 00/15] arm/arm64: add ITS framework
https://lists.gnu.org/archive/html/qemu-devel/2016-12/msg00575.html
[2] [kvm-unit-tests PATCH 00/17] arm: gic: Test SPIs and interrupt groups
https://patchwork.kernel.org/cover/11234975/
[3] [PATCH] KVM: arm/arm64: vgic-its: Fix restoration of unmapped
collections (https://lkml.org/lkml/2019/12/13/205)
Execution:
x For ITS migration testing use:
./run_tests.sh -g migration (block on TCG)
x For other ITS tests:
./run_tests.sh -g its
x non migration tests can be launched invidually. For instance:
./arm-run arm/gic.flat -smp 8 -append 'its-trigger'
Andre Przywara (1):
arm: gic: Provide per-IRQ helper functions
Eric Auger (13):
libcflat: Add other size defines
arm/arm64: gic: Introduce setup_irq() helper arm/arm64: gicv3: Add some re-distributor defines arm/arm64: ITS: Introspection tests
arm/arm64: gicv3: Set the LPI config and pending tables arm/arm64: gicv3: Enable/Disable LPIs at re-distributor level arm/arm64: ITS: its_enable_defaults
arm/arm64: ITS: Device and collection Initialization arm/arm64: ITS: commands
arm/arm64: ITS: INT functional tests
arm/run: Allow Migration tests
arm/arm64: ITS: migration tests
arm/arm64: ITS: pending table migration test
arm/Makefile.arm64 | 1 + arm/Makefile.common | 2 +- arm/gic.c | 488 +++++++++++++++++++++++++++++++++++-- arm/run | 2 +- arm/unittests.cfg | 38 +++ lib/arm/asm/gic-v3-its.h | 167 +++++++++++++ lib/arm/asm/gic-v3.h | 25 ++ lib/arm/asm/gic.h | 8 + lib/arm/gic-v3-its-cmd.c | 454 ++++++++++++++++++++++++++++++++++ lib/arm/gic-v3-its.c | 243 ++++++++++++++++++ lib/arm/gic-v3.c | 81 ++++++ lib/arm/gic.c | 120 ++++++++- lib/arm/io.c | 28 +++ lib/arm64/asm/gic-v3-its.h | 1 + lib/libcflat.h | 3 + 15 files changed, 1633 insertions(+), 28 deletions(-) create mode 100644 lib/arm/asm/gic-v3-its.h create mode 100644 lib/arm/gic-v3-its-cmd.c create mode 100644 lib/arm/gic-v3-its.c create mode 100644 lib/arm64/asm/gic-v3-its.h
Signature · 1 line
2.20.1
PATCH v3 1/14 libcflat: Add other size defines
Introduce additional SZ_256, SZ_8K, SZ_16K macros that will be used by ITS tests.
--- lib/libcflat.h | 3 +++ 1 file changed, 3 insertions(+)
lib/libcflat.h +3 -0
@@ -36,7 +36,10 @@#define ALIGN (x, a) __ALIGN((x), (a))#define IS_ALIGNED (x, a) (((x) & ((typeof(x))(a) - 1)) == 0) +#define SZ_256 (1 << 8)#define SZ_4K (1 << 12) +#define SZ_8K (1 << 13) +#define SZ_16K (1 << 14)#define SZ_64K (1 << 16)#define SZ_2M (1 << 21)#define SZ_1G (1 << 30)
Signature · 1 line
2.20.1
PATCH v3 2/14 arm: gic: Provide per-IRQ helper functions
In reply to Eric AugerFrom: Andre Przywara <[email protected]>
A common theme when accessing per-IRQ parameters in the GIC distributor is to set fields of a certain bit width in a range of MMIO registers. Examples are the enabled status (one bit per IRQ), the level/edge configuration (2 bits per IRQ) or the priority (8 bits per IRQ).
Add a generic helper function which is able to mask and set the respective number of bits, given the IRQ number and the MMIO offset. Provide wrappers using this function to easily allow configuring an IRQ.
For now assume that private IRQ numbers always refer to the current CPU. In a GICv2 accessing the "other" private IRQs is not easily doable (the registers are banked per CPU on the same MMIO address), so we impose the same limitation on GICv3, even though those registers are not banked there anymore.
---
initialize reg
--- lib/arm/asm/gic-v3.h | 2 + lib/arm/asm/gic.h | 8 ++++ lib/arm/gic.c | 90 ++++++++++++++++++++++++++++++++++++++++++++ 3 files changed, 100 insertions(+)
lib/arm/asm/gic-v3.h +2 -0
@@ -23,6 +23,8 @@#define GICD_CTLR_ENABLE_G1A (1U << 1)#define GICD_CTLR_ENABLE_G1 (1U << 0) +#define GICD_IROUTER 0x6000 +/* Re-Distributor registers, offsets from RD_base */ #define GICR_TYPER 0x0008
lib/arm/asm/gic.h +8 -0
@@ -74,5 +74,13 @@ extern void gic_write_eoir(u32 irqstat);extern void gic_ipi_send_single (int irq,int cpu);extern void gic_ipi_send_mask (int irq,const cpumask_t *dest); +void gic_set_irq_bit (int irq,int offset); +void gic_enable_irq (int irq); +void gic_disable_irq (int irq); +void gic_set_irq_priority (int irq,u8 prio); +void gic_set_irq_target (int irq,int cpu); +void gic_set_irq_group (int irq,int group); +int gic_get_irq_group (int irq); +#endif /* !__ASSEMBLY__ */ #endif /* _ASMARM_GIC_H_ */
lib/arm/gic.c +90 -0
93 lines not on this page · raw message
Signature · 1 line
2.20.1
PATCH v3 3/14 arm/arm64: gic: Introduce setup_irq() helper
In reply to Eric Augeripi_enable() code would be reusable for other interrupts than IPI. Let's rename it setup_irq() and pass an interrupt handler pointer.
---
v2 -> v3:
- do not export setup_irq anymore
--- arm/gic.c | 20 +++++++------------- 1 file changed, 7 insertions(+), 13 deletions(-)
arm/gic.c +7 -13
@@ -34,6 +34,7 @@ static struct gic *gic;static int acked[NR_CPUS ], spurious[NR_CPUS ];static int bad_sender[NR_CPUS ], bad_irq[NR_CPUS ];static cpumask_t ready; +typedef void (*handler_t)(struct pt_regs *regs __unused);static void nr_cpu_check (int nr) {
@@ -215,20 +216,20 @@ static void ipi_test_smp(void)report_prefix_pop (); } -static void ipi_enable (void ) +static void setup_irq (handler_t handler) {gic_enable_defaults ();#ifdef __arm__ -install_exception_handler (EXCPTN_IRQ , ipi_handler); +install_exception_handler (EXCPTN_IRQ , handler);#else -install_irq_handler (EL1H_IRQ , ipi_handler); +install_irq_handler (EL1H_IRQ , handler);#endif local_irq_enable (); }static void ipi_send (void ) { -ipi_enable (); +setup_irq (ipi_handler);wait_on_ready ();ipi_test_self ();ipi_test_smp ();
@@ -238,7 +239,7 @@ static void ipi_send(void)static void ipi_recv (void ) { -ipi_enable (); +setup_irq (ipi_handler);cpumask_set_cpu (smp_processor_id (), &ready);while (1 )wfi ();
@@ -295,14 +296,7 @@ static void ipi_clear_active_handler(struct pt_regs *regs __unused)static void run_active_clear_test (void ) {report_prefix_push ("active" ); -gic_enable_defaults (); -#ifdef __arm__ -install_exception_handler (EXCPTN_IRQ , ipi_clear_active_handler); -#else -install_irq_handler (EL1H_IRQ , ipi_clear_active_handler); -#endif -local_irq_enable (); - +setup_irq (ipi_clear_active_handler);ipi_test_self ();report_prefix_pop (); }
Signature · 1 line
2.20.1
PATCH v3 4/14 arm/arm64: gicv3: Add some re-distributor defines
In reply to Eric AugerPROPBASER, PENDBASE and GICR_CTRL will be used for LPI management.
--- lib/arm/asm/gic-v3.h | 6 ++++++ 1 file changed, 6 insertions(+)
lib/arm/asm/gic-v3.h +6 -0
@@ -18,6 +18,7 @@* We expect to be run in Non-secure mode, thus we define the * group1 enable bits with respect to that view. */ +#define GICD_CTLR 0x0000#define GICD_CTLR_RWP (1U << 31)#define GICD_CTLR_ARE_NS (1U << 4)#define GICD_CTLR_ENABLE_G1A (1U << 1)
@@ -36,6 +37,11 @@#define GICR_ICENABLER0 GICD_ICENABLER#define GICR_IPRIORITYR0 GICD_IPRIORITYR +#define GICR_PROPBASER 0x0070 +#define GICR_PENDBASER 0x0078 +#define GICR_CTLR GICD_CTLR +#define GICR_CTLR_ENABLE_LPIS (1UL << 0) +#define ICC_SGI1R_AFFINITY_1_SHIFT 16#define ICC_SGI1R_AFFINITY_2_SHIFT 32#define ICC_SGI1R_AFFINITY_3_SHIFT 48
Signature · 1 line
2.20.1
Re: [kvm-unit-tests PATCH v3 04/14] arm/arm64: gicv3: Add some re-distributor defines
Hi Eric,
Quoted 20 lines · Eric Auger
PROPBASER, PENDBASE and
GICR_CTRLwill be used for LPI management.Signed-off-by Eric Auger <[email protected]>--- lib/arm/asm/gic-v3.h | 6 ++++++ 1 file changed, 6 insertions(+)lib/arm/asm/gic-v3.h +1
-0@@ -18,6 +18,7 @@* We expect to be run in Non-secure mode, thus we define the * group1 enable bits with respect to that view. */ +#define GICD_CTLR 0x0000#define GICD_CTLR_RWP (1U << 31)#define GICD_CTLR_ARE_NS (1U << 4)#define GICD_CTLR_ENABLE_G1A (1U << 1)
lib/arm/asm/gic-v3.h line 41
Quoted 5 lines
lib/arm/asm/gic-v3.h +2
-0@@ -36,6 +37,11 @@#define GICR_ICENABLER0 GICD_ICENABLER#define GICR_IPRIORITYR0 GICD_IPRIORITYR +#define GICR_PROPBASER 0x0070 +#define GICR_PENDBASER 0x0078
nit: we'd better use tab instead of space.
Quoted 7 lines
lib/arm/asm/gic-v3.h +3
-0+#define GICR_CTLR GICD_CTLR +#define GICR_CTLR_ENABLE_LPIS (1UL << 0) +#define ICC_SGI1R_AFFINITY_1_SHIFT 16#define ICC_SGI1R_AFFINITY_2_SHIFT 32#define ICC_SGI1R_AFFINITY_3_SHIFT 48
PATCH v3 5/14 arm/arm64: ITS: Introspection tests
In reply to Eric AugerDetect the presence of an ITS as part of the GICv3 init routine, initialize its base address and read few registers the IIDR, the TYPER to store its dimensioning parameters. Also parse the BASER registers.
This is our first ITS test, belonging to a new "its" group.
---
v2 -> v3:
- updated dates and changed author
- squash "arm/arm64: ITS: Test BASER" into this patch but removes setup_baser which will be introduced later.
- only compile on aarch64
- restrict the new test to aarch64
v1 -> v2:
- clean GITS_TYPER macros and unused fields in typer struct
- remove memory attribute related macros
- remove everything related to memory attributes
- s/dev_baser/coll_baser/ in report_info
- add extra line
- removed index filed in its_baser
--- arm/Makefile.arm64 | 1 + arm/gic.c | 49 ++++++++++++++++++ arm/unittests.cfg | 7 +++ lib/arm/asm/gic-v3-its.h | 103 +++++++++++++++++++++++++++++++++++++ lib/arm/gic-v3-its.c | 88 +++++++++++++++++++++++++++++++ lib/arm/gic.c | 30 +++++++++-- lib/arm64/asm/gic-v3-its.h | 1 + 7 files changed, 274 insertions(+), 5 deletions(-) create mode 100644 lib/arm/asm/gic-v3-its.h create mode 100644 lib/arm/gic-v3-its.c create mode 100644 lib/arm64/asm/gic-v3-its.h
arm/Makefile.arm64 +1 -0
@@ -19,6 +19,7 @@ endef cstart.o = $(TEST_DIR)/cstart64.o cflatobjs += lib/arm64/processor.o cflatobjs += lib/arm64/spinlock.o +cflatobjs += lib/arm/gic-v3-its.o OBJDIRS += lib/arm64
arm/gic.c +49 -0
67 lines not on this page · raw message
arm/unittests.cfg +7 -0
@@ -122,6 +122,13 @@ smp = $MAX_SMP extra_params = -machine gic-version=3 -append 'active' groups = gic +[its-introspection] +file = gic.flat +smp = $MAX_SMP +extra_params = -machine gic-version=3 -append 'its-introspection' +groups = its +arch = arm64 + # Test PSCI emulation [psci] file = psci.flat
lib/arm/asm/gic-v3-its.h new file +103 -0
103 lines not on this page · raw message
lib/arm/gic-v3-its.c new file +88 -0
88 lines not on this page · raw message
lib/arm/gic.c +25 -5
@@ -6,9 +6,11 @@#include <devicetree.h> #include <asm/gic.h> #include <asm/io.h> +#include <asm/gic-v3-its.h> struct gicv2_data gicv2_data;struct gicv3_data gicv3_data; +struct its_data its_data;struct gic_common_ops {void (*enable_defaults)(void );
@@ -44,12 +46,13 @@ static const struct gic_common_ops gicv3_common_ops = {* Documentation/devicetree/bindings/interrupt-controller/arm,gic-v3.txt */ static bool -gic_get_dt_bases (const char *compatible,void **base1,void **base2) +gic_get_dt_bases (const char *compatible,void **base1,void **base2,void **base3) {struct dt_pbus_reg reg; -struct dt_device gic; +struct dt_device gic, its;struct dt_bus bus; -int node, ret, i; +int node, subnode, ret, i, len; +const void *fdt =dt_fdt ();dt_bus_init_defaults (&bus);dt_device_init (&gic, &bus,NULL );
@@ -74,19 +77,35 @@ gic_get_dt_bases(const char *compatible, void **base1, void **base2) base2[i] =ioremap (reg.addr, reg.size); } +if (base3 && !strcmp (compatible,"arm,gic-v3" )) { +dt_for_each_subnode (node, subnode) { +const struct fdt_property *prop; + + prop =fdt_get_property (fdt, subnode,"compatible" , &len); +if (!strcmp ((char *)prop->data,"arm,gic-v3-its" )) { +dt_device_bind_node (&its, subnode); + ret =dt_pbus_translate (&its,0 , ®); +assert (ret ==0 ); + *base3 =ioremap (reg.addr, reg.size); +break ; + } + } + + } +return true; }int gicv2_init (void ) {return gic_get_dt_bases ("arm,cortex-a15-gic" , - &gicv2_data.dist_base, &gicv2_data.cpu_base); + &gicv2_data.dist_base, &gicv2_data.cpu_base,NULL ); }int gicv3_init (void ) {return gic_get_dt_bases ("arm,gic-v3" , &gicv3_data.dist_base, - &gicv3_data.redist_bases[0 ]); + &gicv3_data.redist_bases[0 ], &its_data.base); }int gic_version (void )
@@ -104,6 +123,7 @@ int gic_init(void) gic_common_ops = &gicv2_common_ops; else if (gicv3_init()) gic_common_ops = &gicv3_common_ops; +its_init ();return gic_version (); }
lib/arm64/asm/gic-v3-its.h new file +1 -0
@@ -0,0 +1 @@ +#include "../../arm/asm/gic-v3-its.h"
Signature · 1 line
2.20.1
Zenghui Yu
reviewerRe: [kvm-unit-tests PATCH v3 05/14] arm/arm64: ITS: Introspection tests
Hi Eric,
Quoted 90 lines · Eric Auger
Detect the presence of an ITS as part of the GICv3 init routine, initialize its base address and read few registers the IIDR, the TYPER to store its dimensioning parameters. Also parse the BASER registers.
This is our first ITS test, belonging to a new "its" group.
Signed-off-by Eric Auger <[email protected]>---
v2 -> v3:
- updated dates and changed author
- squash "arm/arm64: ITS: Test BASER" into this patch but removessetup_baserwhich will be introduced later.
- only compile on aarch64
- restrict the new test to aarch64v1 -> v2:
- cleanGITS_TYPERmacros and unused fields in typer struct
- remove memory attribute related macros
- remove everything related to memory attributes
- s/dev_baser/coll_baser/ inreport_info
- add extra line
- removed index filed inits_baser--- arm/Makefile.arm64 | 1 + arm/gic.c | 49 ++++++++++++++++++ arm/unittests.cfg | 7 +++ lib/arm/asm/gic-v3-its.h | 103 +++++++++++++++++++++++++++++++++++++ lib/arm/gic-v3-its.c | 88 +++++++++++++++++++++++++++++++ lib/arm/gic.c | 30 +++++++++-- lib/arm64/asm/gic-v3-its.h | 1 + 7 files changed, 274 insertions(+), 5 deletions(-)create mode 100644
lib/arm/asm/gic-v3-its.h
create mode 100644lib/arm/gic-v3-its.c
create mode 100644lib/arm64/asm/gic-v3-its.harm/Makefile.arm64 +1
-0@@ -19,6 +19,7 @@ endef cstart.o = $(TEST_DIR)/cstart64.o cflatobjs += lib/arm64/processor.o cflatobjs += lib/arm64/spinlock.o +cflatobjs += lib/arm/gic-v3-its.o OBJDIRS += lib/arm64arm/gic.c +25
-0@@ -16,6 +16,7 @@#include <asm/processor.h> #include <asm/delay.h> #include <asm/gic.h> +#include <asm/gic-v3-its.h> #include <asm/smp.h> #include <asm/barrier.h> #include <asm/io.h> @@ -518,6 +519,50 @@ static void gic_test_mmio(void)test_targets (nr_irqs); } +#if defined(__arm__) + +static void test_its_introspection (void ) {} + +#else /* __arch64__ */ + +static void test_its_introspection (void ) +{ +struct its_baser *dev_baser, *coll_baser; +struct its_typer *typer = &its_data.typer; + +if (!gicv3_its_base ()) { +report_skip ("No ITS, skip ..." ); +return ; + } + +/* IIDR */ +report (test_readonly_32 (gicv3_its_base () +GITS_IIDR , false), +"GITS_IIDR is read-only" ), + +/* TYPER */ +report (test_readonly_32 (gicv3_its_base () +GITS_TYPER , false), +"GITS_TYPER is read-only" ); +
arm/gic.c static void gic_test_mmio(void) · line 555
Quoted 10 lines
arm/gic.c +10
-0+report (typer->phys_lpi,"ITS supports physical LPIs" ); +report_info ("vLPI support: %s" , typer->virt_lpi ?"yes" :"no" ); +report_info ("ITT entry size = 0x%x" , typer->ite_size); +report_info ("Bit Count: EventID=%d DeviceId=%d CollId=%d" , + typer->eventid_bits, typer->deviceid_bits, + typer->collid_bits); +report (typer->eventid_bits && typer->deviceid_bits && + typer->collid_bits,"ID spaces" ); +report_info ("Target address format %s" , + typer->pta ?"Redist basse address" :"PE #" );
typo: s/basse/base/
arm/gic.c static void gic_test_mmio(void) · line 561
Quoted 6 lines
arm/gic.c +6
-0+ + dev_baser =its_lookup_baser (GITS_BASER_TYPE_DEVICE ); + coll_baser =its_lookup_baser (GITS_BASER_TYPE_COLLECTION ); +report (dev_baser && coll_baser,"detect device and collection BASER" ); +report_info ("device baser entry_size = 0x%x" , dev_baser->esz); +report_info ("collection baser entry_size = 0x%x" , coll_baser->esz);
How about "device table entry_size = ..." and "collection table entry_size = ..."?
Quoted 181 lines
arm/gic.c +8
-0+} + +#endif +int main (int argc,char **argv) {if (!gic_init ()) {@@ -549,6 +594,10 @@ int main(int argc, char **argv)report_prefix_push (argv[1 ]);gic_test_mmio ();report_prefix_pop (); + }else if (strcmp (argv[1 ],"its-introspection" ) ==0 ) { +report_prefix_push (argv[1 ]); +test_its_introspection (); +report_prefix_pop (); }else {report_abort ("Unknown subtest '%s'" , argv[1 ]); }arm/unittests.cfg +7
-0@@ -122,6 +122,13 @@ smp = $MAX_SMP extra_params = -machine gic-version=3 -append 'active' groups = gic +[its-introspection] +file = gic.flat +smp = $MAX_SMP +extra_params = -machine gic-version=3 -append 'its-introspection' +groups = its +arch = arm64 + # Test PSCI emulation [psci] file = psci.flatlib/arm/asm/gic-v3-its.h new file +103
-0@@ -0,0 +1,103 @@ +/* +* All ITS* defines are lifted from include/linux/irqchip/arm-gic-v3.h +* +* Copyright (C) 2020, Red Hat Inc, Eric Auger <[email protected]> +* +* This work is licensed under the terms of the GNU LGPL, version 2. +*/ +#ifndef _ASMARM_GIC_V3_ITS_H_ +#define _ASMARM_GIC_V3_ITS_H_ + +#ifndef __ASSEMBLY__ + +struct its_typer { +unsigned int ite_size; +unsigned int eventid_bits; +unsigned int deviceid_bits; +unsigned int collid_bits; +bool pta; +bool phys_lpi; +bool virt_lpi; +}; + +struct its_baser { +int type; +size_t psz; +int nr_pages; +bool indirect; +phys_addr_t table_addr; +bool valid; +int esz; +}; + +#define GITS_BASER_NR_REGS 8 + +struct its_data { +void *base; +struct its_typer typer; +struct its_baser baser[GITS_BASER_NR_REGS ]; +}; + +extern struct its_data its_data; + +#define gicv3_its_base () (its_data.base) + +#if defined(__aarch64__) + +#define GITS_CTLR 0x0000 +#define GITS_IIDR 0x0004 +#define GITS_TYPER 0x0008 +#define GITS_CBASER 0x0080 +#define GITS_CWRITER 0x0088 +#define GITS_CREADR 0x0090 +#define GITS_BASER 0x0100 + +#define GITS_TYPER_PLPIS BIT(0) +#define GITS_TYPER_VLPIS BIT(1) +#define GITS_TYPER_ITT_ENTRY_SIZE GENMASK_ULL(7, 4) +#define GITS_TYPER_ITT_ENTRY_SIZE_SHIFT 4 +#define GITS_TYPER_IDBITS GENMASK_ULL(8, 12) +#define GITS_TYPER_IDBITS_SHIFT 8 +#define GITS_TYPER_DEVBITS GENMASK_ULL(13, 17) +#define GITS_TYPER_DEVBITS_SHIFT 13 +#define GITS_TYPER_PTA BIT(19) +#define GITS_TYPER_CIDBITS GENMASK_ULL(32, 35) +#define GITS_TYPER_CIDBITS_SHIFT 32 +#define GITS_TYPER_CIL BIT(36) + +#define GITS_CTLR_ENABLE (1U << 0) + +#define GITS_CBASER_VALID (1UL << 63) + +#define GITS_BASER_VALID BIT(63) +#define GITS_BASER_INDIRECT BIT(62) +#define GITS_BASER_TYPE_SHIFT (56) +#define GITS_BASER_TYPE (r) (((r) >> GITS_BASER_TYPE_SHIFT) & 7) +#define GITS_BASER_ENTRY_SIZE_SHIFT (48) +#define GITS_BASER_ENTRY_SIZE (r) ((((r) >> GITS_BASER_ENTRY_SIZE_SHIFT) & 0x1f) + 1) +#define GITS_BASER_PAGE_SIZE_SHIFT (8) +#define GITS_BASER_PAGE_SIZE_4K (0UL << GITS_BASER_PAGE_SIZE_SHIFT) +#define GITS_BASER_PAGE_SIZE_16K (1UL << GITS_BASER_PAGE_SIZE_SHIFT) +#define GITS_BASER_PAGE_SIZE_64K (2UL << GITS_BASER_PAGE_SIZE_SHIFT) +#define GITS_BASER_PAGE_SIZE_MASK (3UL << GITS_BASER_PAGE_SIZE_SHIFT) +#define GITS_BASER_PAGES_MAX 256 +#define GITS_BASER_PAGES_SHIFT (0) +#define GITS_BASER_NR_PAGES (r) (((r) & 0xff) + 1) +#define GITS_BASER_PHYS_ADDR_MASK 0xFFFFFFFFF000 +#define GITS_BASER_TYPE_NONE 0 +#define GITS_BASER_TYPE_DEVICE 1 +#define GITS_BASER_TYPE_COLLECTION 4 + +extern void its_parse_typer (void ); +extern void its_init (void ); +extern int its_parse_baser (int i,struct its_baser *baser); +extern struct its_baser *its_lookup_baser (int type); + +#else /* __arm__ */ + +static inline void its_init (void ) {} + +#endif + +#endif /* !__ASSEMBLY__ */ +#endif /* _ASMARM_GIC_V3_ITS_H_ */ lib/arm/gic-v3-its.c new file +30
-0@@ -0,0 +1,88 @@ +/* +* Copyright (C) 2020, Red Hat Inc, Eric Auger <[email protected]> +* +* This work is licensed under the terms of the GNU LGPL, version 2. +*/ +#include <asm/gic.h> +#include <alloc_page.h> +#include <asm/gic-v3-its.h> + +void its_parse_typer (void ) +{ +u64 typer =readq (gicv3_its_base () +GITS_TYPER ); + + its_data.typer.ite_size = ((typer &GITS_TYPER_ITT_ENTRY_SIZE ) >> +GITS_TYPER_ITT_ENTRY_SIZE_SHIFT ) +1 ; + its_data.typer.pta = typer &GITS_TYPER_PTA ; + its_data.typer.eventid_bits = ((typer &GITS_TYPER_IDBITS ) >> +GITS_TYPER_IDBITS_SHIFT ) +1 ; + its_data.typer.deviceid_bits = ((typer &GITS_TYPER_DEVBITS ) >> +GITS_TYPER_DEVBITS_SHIFT ) +1 ; + +if (typer &GITS_TYPER_CIL ) + its_data.typer.collid_bits = ((typer &GITS_TYPER_CIDBITS ) >> +GITS_TYPER_CIDBITS_SHIFT ) +1 ; +else + its_data.typer.collid_bits =16 ; + + its_data.typer.virt_lpi = typer &GITS_TYPER_VLPIS ; + its_data.typer.phys_lpi = typer &GITS_TYPER_PLPIS ; +}
lib/arm/gic-v3-its.c line 40
Quoted 10 lines
lib/arm/gic-v3-its.c new file +10
-0+ +int its_parse_baser (int i,struct its_baser *baser) +{ +void *reg_addr =gicv3_its_base () +GITS_BASER + i *8 ; +u64 val =readq (reg_addr); + +if (!val) { +memset (baser,0 ,sizeof (*baser)); +return -1 ; + }
Unimplemented BASERs? How about using something like:
if (GITS_BASER_TYPE(val) == GITS_BASER_TYPE_NONE) {
[...]
}to make it a bit more explicit?
But feel free to just ignore it because you're right that the unimplemented BASERs are RES0.
Thanks,
Zenghui
Quoted 135 lines
lib/arm/gic-v3-its.c new file +48
-0+ + baser->valid = val &GITS_BASER_VALID ; + baser->indirect = val &GITS_BASER_INDIRECT ; + baser->type =GITS_BASER_TYPE (val); + baser->esz =GITS_BASER_ENTRY_SIZE (val); + baser->nr_pages =GITS_BASER_NR_PAGES (val); + baser->table_addr = val &GITS_BASER_PHYS_ADDR_MASK ; +switch (val &GITS_BASER_PAGE_SIZE_MASK ) { +case GITS_BASER_PAGE_SIZE_4K : + baser->psz =SZ_4K ; +break ; +case GITS_BASER_PAGE_SIZE_16K : + baser->psz =SZ_16K ; +break ; +case GITS_BASER_PAGE_SIZE_64K : + baser->psz =SZ_64K ; +break ; +default : + baser->psz =SZ_64K ; + } +return 0 ; +} + +struct its_baser *its_lookup_baser (int type) +{ +int i; + +for (i =0 ; i <GITS_BASER_NR_REGS ; i++) { +struct its_baser *baser = &its_data.baser[i]; + +if (baser->type == type) +return baser; + } +return NULL ; +} + +void its_init (void ) +{ +int i; + +if (!its_data.base) +return ; + +its_parse_typer (); +for (i =0 ; i <GITS_BASER_NR_REGS ; i++) +its_parse_baser (i, &its_data.baser[i]); +} +lib/arm/gic.c +25
-5@@ -6,9 +6,11 @@#include <devicetree.h> #include <asm/gic.h> #include <asm/io.h> +#include <asm/gic-v3-its.h> struct gicv2_data gicv2_data;struct gicv3_data gicv3_data; +struct its_data its_data;struct gic_common_ops {void (*enable_defaults)(void );@@ -44,12 +46,13 @@ static const struct gic_common_ops gicv3_common_ops = {* Documentation/devicetree/bindings/interrupt-controller/arm,gic-v3.txt */ static bool -gic_get_dt_bases (const char *compatible,void **base1,void **base2) +gic_get_dt_bases (const char *compatible,void **base1,void **base2,void **base3) {struct dt_pbus_reg reg; -struct dt_device gic; +struct dt_device gic, its;struct dt_bus bus; -int node, ret, i; +int node, subnode, ret, i, len; +const void *fdt =dt_fdt ();dt_bus_init_defaults (&bus);dt_device_init (&gic, &bus,NULL );@@ -74,19 +77,35 @@ gic_get_dt_bases(const char *compatible, void **base1, void **base2) base2[i] =ioremap (reg.addr, reg.size); } +if (base3 && !strcmp (compatible,"arm,gic-v3" )) { +dt_for_each_subnode (node, subnode) { +const struct fdt_property *prop; + + prop =fdt_get_property (fdt, subnode,"compatible" , &len); +if (!strcmp ((char *)prop->data,"arm,gic-v3-its" )) { +dt_device_bind_node (&its, subnode); + ret =dt_pbus_translate (&its,0 , ®); +assert (ret ==0 ); + *base3 =ioremap (reg.addr, reg.size); +break ; + } + } + + } +return true; }int gicv2_init (void ) {return gic_get_dt_bases ("arm,cortex-a15-gic" , - &gicv2_data.dist_base, &gicv2_data.cpu_base); + &gicv2_data.dist_base, &gicv2_data.cpu_base,NULL ); }int gicv3_init (void ) {return gic_get_dt_bases ("arm,gic-v3" , &gicv3_data.dist_base, - &gicv3_data.redist_bases[0 ]); + &gicv3_data.redist_bases[0 ], &its_data.base); }int gic_version (void )@@ -104,6 +123,7 @@ int gic_init(void) gic_common_ops = &gicv2_common_ops; else if (gicv3_init()) gic_common_ops = &gicv3_common_ops; +its_init ();return gic_version (); }lib/arm64/asm/gic-v3-its.h new file +1
-0@@ -0,0 +1 @@ +#include "../../arm/asm/gic-v3-its.h"
Re: [kvm-unit-tests PATCH v3 05/14] arm/arm64: ITS: Introspection tests
In reply to Eric AugerQuoted 43 lines · Eric Auger
Detect the presence of an ITS as part of the GICv3 init routine, initialize its base address and read few registers the IIDR, the TYPER to store its dimensioning parameters. Also parse the BASER registers.
This is our first ITS test, belonging to a new "its" group.
Signed-off-by Eric Auger <[email protected]>---
v2 -> v3:
- updated dates and changed author
- squash "arm/arm64: ITS: Test BASER" into this patch but removessetup_baserwhich will be introduced later.
- only compile on aarch64
- restrict the new test to aarch64v1 -> v2:
- cleanGITS_TYPERmacros and unused fields in typer struct
- remove memory attribute related macros
- remove everything related to memory attributes
- s/dev_baser/coll_baser/ inreport_info
- add extra line
- removed index filed inits_baser--- arm/Makefile.arm64 | 1 + arm/gic.c | 49 ++++++++++++++++++ arm/unittests.cfg | 7 +++ lib/arm/asm/gic-v3-its.h | 103 +++++++++++++++++++++++++++++++++++++ lib/arm/gic-v3-its.c | 88 +++++++++++++++++++++++++++++++ lib/arm/gic.c | 30 +++++++++-- lib/arm64/asm/gic-v3-its.h | 1 + 7 files changed, 274 insertions(+), 5 deletions(-) create mode 100644 lib/arm/asm/gic-v3-its.h create mode 100644 lib/arm/gic-v3-its.c create mode 100644 lib/arm64/asm/gic-v3-its.harm/Makefile.arm64 +0
-0
arm/Makefile.arm64 endef · line 22
Quoted 4 lines
arm/Makefile.arm64 +1
-0@@ -19,6 +19,7 @@ endef cstart.o = $(TEST_DIR)/cstart64.o cflatobjs += lib/arm64/processor.o cflatobjs += lib/arm64/spinlock.o +cflatobjs += lib/arm/gic-v3-its.o
If gic-v3-its.c will never be compiled for arm, then it should probably go lib/arm64, not lib/arm. Same comment for all other new source and header files. The only problem with that is...
arm/gic.c line 19
Quoted 12 lines
arm/Makefile.arm64 +0
-0OBJDIRS += lib/arm64arm/gic.c +1
-0@@ -16,6 +16,7 @@#include <asm/processor.h> #include <asm/delay.h> #include <asm/gic.h> +#include <asm/gic-v3-its.h>
...here where we'd have to do
#if defined(__aarch64__)
#include <asm/gic-v3-its.h>
#endif
which is ugly. However that can be avoided by adding an empty lib/arm/asm/gic-v3-its.h file.
arm/gic.c static void gic_test_mmio(void) · line 524
Quoted 10 lines
arm/gic.c +3
-0#include <asm/smp.h> #include <asm/barrier.h> #include <asm/io.h> @@ -518,6 +519,50 @@ static void gic_test_mmio(void)test_targets (nr_irqs); } +#if defined(__arm__) + +static void test_its_introspection (void ) {}
static void test_its_introspection(void)
{
report_abort(...);}
Quoted 80 lines
arm/gic.c +45
-0+ +#else /* __arch64__ */ + +static void test_its_introspection (void ) +{ +struct its_baser *dev_baser, *coll_baser; +struct its_typer *typer = &its_data.typer; + +if (!gicv3_its_base ()) { +report_skip ("No ITS, skip ..." ); +return ; + } + +/* IIDR */ +report (test_readonly_32 (gicv3_its_base () +GITS_IIDR , false), +"GITS_IIDR is read-only" ), + +/* TYPER */ +report (test_readonly_32 (gicv3_its_base () +GITS_TYPER , false), +"GITS_TYPER is read-only" ); + +report (typer->phys_lpi,"ITS supports physical LPIs" ); +report_info ("vLPI support: %s" , typer->virt_lpi ?"yes" :"no" ); +report_info ("ITT entry size = 0x%x" , typer->ite_size); +report_info ("Bit Count: EventID=%d DeviceId=%d CollId=%d" , + typer->eventid_bits, typer->deviceid_bits, + typer->collid_bits); +report (typer->eventid_bits && typer->deviceid_bits && + typer->collid_bits,"ID spaces" ); +report_info ("Target address format %s" , + typer->pta ?"Redist basse address" :"PE #" ); + + dev_baser =its_lookup_baser (GITS_BASER_TYPE_DEVICE ); + coll_baser =its_lookup_baser (GITS_BASER_TYPE_COLLECTION ); +report (dev_baser && coll_baser,"detect device and collection BASER" ); +report_info ("device baser entry_size = 0x%x" , dev_baser->esz); +report_info ("collection baser entry_size = 0x%x" , coll_baser->esz); +} + +#endif +int main (int argc,char **argv) {if (!gic_init ()) {@@ -549,6 +594,10 @@ int main(int argc, char **argv)report_prefix_push (argv[1 ]);gic_test_mmio ();report_prefix_pop (); + }else if (strcmp (argv[1 ],"its-introspection" ) ==0 ) { +report_prefix_push (argv[1 ]); +test_its_introspection (); +report_prefix_pop (); }else {report_abort ("Unknown subtest '%s'" , argv[1 ]); }arm/unittests.cfg +7
-0@@ -122,6 +122,13 @@ smp = $MAX_SMP extra_params = -machine gic-version=3 -append 'active' groups = gic +[its-introspection] +file = gic.flat +smp = $MAX_SMP +extra_params = -machine gic-version=3 -append 'its-introspection' +groups = its +arch = arm64 + # Test PSCI emulation [psci] file = psci.flatlib/arm/asm/gic-v3-its.h new file +1
-0@@ -0,0 +1,103 @@ +/*
lib/arm/asm/gic-v3-its.h line 11
Quoted 10 lines
lib/arm/asm/gic-v3-its.h new file +10
-0+* All ITS* defines are lifted from include/linux/irqchip/arm-gic-v3.h +* +* Copyright (C) 2020, Red Hat Inc, Eric Auger <[email protected]> +* +* This work is licensed under the terms of the GNU LGPL, version 2. +*/ +#ifndef _ASMARM_GIC_V3_ITS_H_ +#define _ASMARM_GIC_V3_ITS_H_ + +#ifndef __ASSEMBLY__
Doesn't look like you use the #else /* __ASSEMBLY__ */ side of this. I'd leave out the #ifndef until we add defines we need to access from assembler.
Quoted 77 lines
lib/arm/asm/gic-v3-its.h new file +77
-0+ +struct its_typer { +unsigned int ite_size; +unsigned int eventid_bits; +unsigned int deviceid_bits; +unsigned int collid_bits; +bool pta; +bool phys_lpi; +bool virt_lpi; +}; + +struct its_baser { +int type; +size_t psz; +int nr_pages; +bool indirect; +phys_addr_t table_addr; +bool valid; +int esz; +}; + +#define GITS_BASER_NR_REGS 8 + +struct its_data { +void *base; +struct its_typer typer; +struct its_baser baser[GITS_BASER_NR_REGS ]; +}; + +extern struct its_data its_data; + +#define gicv3_its_base () (its_data.base) + +#if defined(__aarch64__) + +#define GITS_CTLR 0x0000 +#define GITS_IIDR 0x0004 +#define GITS_TYPER 0x0008 +#define GITS_CBASER 0x0080 +#define GITS_CWRITER 0x0088 +#define GITS_CREADR 0x0090 +#define GITS_BASER 0x0100 + +#define GITS_TYPER_PLPIS BIT(0) +#define GITS_TYPER_VLPIS BIT(1) +#define GITS_TYPER_ITT_ENTRY_SIZE GENMASK_ULL(7, 4) +#define GITS_TYPER_ITT_ENTRY_SIZE_SHIFT 4 +#define GITS_TYPER_IDBITS GENMASK_ULL(8, 12) +#define GITS_TYPER_IDBITS_SHIFT 8 +#define GITS_TYPER_DEVBITS GENMASK_ULL(13, 17) +#define GITS_TYPER_DEVBITS_SHIFT 13 +#define GITS_TYPER_PTA BIT(19) +#define GITS_TYPER_CIDBITS GENMASK_ULL(32, 35) +#define GITS_TYPER_CIDBITS_SHIFT 32 +#define GITS_TYPER_CIL BIT(36) + +#define GITS_CTLR_ENABLE (1U << 0) + +#define GITS_CBASER_VALID (1UL << 63) + +#define GITS_BASER_VALID BIT(63) +#define GITS_BASER_INDIRECT BIT(62) +#define GITS_BASER_TYPE_SHIFT (56) +#define GITS_BASER_TYPE (r) (((r) >> GITS_BASER_TYPE_SHIFT) & 7) +#define GITS_BASER_ENTRY_SIZE_SHIFT (48) +#define GITS_BASER_ENTRY_SIZE (r) ((((r) >> GITS_BASER_ENTRY_SIZE_SHIFT) & 0x1f) + 1) +#define GITS_BASER_PAGE_SIZE_SHIFT (8) +#define GITS_BASER_PAGE_SIZE_4K (0UL << GITS_BASER_PAGE_SIZE_SHIFT) +#define GITS_BASER_PAGE_SIZE_16K (1UL << GITS_BASER_PAGE_SIZE_SHIFT) +#define GITS_BASER_PAGE_SIZE_64K (2UL << GITS_BASER_PAGE_SIZE_SHIFT) +#define GITS_BASER_PAGE_SIZE_MASK (3UL << GITS_BASER_PAGE_SIZE_SHIFT) +#define GITS_BASER_PAGES_MAX 256 +#define GITS_BASER_PAGES_SHIFT (0) +#define GITS_BASER_NR_PAGES (r) (((r) & 0xff) + 1) +#define GITS_BASER_PHYS_ADDR_MASK 0xFFFFFFFFF000 +#define GITS_BASER_TYPE_NONE 0 +#define GITS_BASER_TYPE_DEVICE 1
lib/arm/asm/gic-v3-its.h line 98
Quoted 10 lines
lib/arm/asm/gic-v3-its.h new file +10
-0+#define GITS_BASER_TYPE_COLLECTION 4 + +extern void its_parse_typer (void ); +extern void its_init (void ); +extern int its_parse_baser (int i,struct its_baser *baser); +extern struct its_baser *its_lookup_baser (int type); + +#else /* __arm__ */ + +static inline void its_init (void ) {}
Looks like the empty gic-v3-its.h I suggested creating above will actually be useful. We can add stubs like this in it.
Quoted 25 lines
lib/arm/asm/gic-v3-its.h new file +5
-0+ +#endif + +#endif /* !__ASSEMBLY__ */ +#endif /* _ASMARM_GIC_V3_ITS_H_ */ lib/arm/gic-v3-its.c new file +14
-0@@ -0,0 +1,88 @@ +/* +* Copyright (C) 2020, Red Hat Inc, Eric Auger <[email protected]> +* +* This work is licensed under the terms of the GNU LGPL, version 2. +*/ +#include <asm/gic.h> +#include <alloc_page.h> +#include <asm/gic-v3-its.h> + +void its_parse_typer (void ) +{ +u64 typer =readq (gicv3_its_base () +GITS_TYPER ); + + its_data.typer.ite_size = ((typer & GITS_TYPER_ITT_ENTRY_SIZE) >>
lib/arm/gic-v3-its.c line 24
Quoted 10 lines
lib/arm/gic-v3-its.c new file +10
-0+ GITS_TYPER_ITT_ENTRY_SIZE_SHIFT) + 1; + its_data.typer.pta = typer &GITS_TYPER_PTA ; + its_data.typer.eventid_bits = ((typer &GITS_TYPER_IDBITS ) >> +GITS_TYPER_IDBITS_SHIFT ) +1 ; + its_data.typer.deviceid_bits = ((typer &GITS_TYPER_DEVBITS ) >> +GITS_TYPER_DEVBITS_SHIFT ) +1 ; + +if (typer &GITS_TYPER_CIL ) + its_data.typer.collid_bits = ((typer &GITS_TYPER_CIDBITS ) >> +GITS_TYPER_CIDBITS_SHIFT ) +1 ;
nit: please consider aligning like this
((typer & MASK) >>
SHIFT) + 1;
Or, maybe better to macro it
#define TYPER_FIELD(typer, mask, shift) (((type) & (mask) >> (shift)) + 1)
And, rather than have a bunch of 'its_data.typer's we could use an alias, helping us stay within 120 chars.
struct its_typer *t = &its_data.typer;
t->ite_size = TYPER_FIELD(typer, GITS_TYPER_ITT_ENTRY_SIZE,
GITS_TYPER_ITT_ENTRY_SIZE_SHIFT);
Quoted 98 lines
lib/arm/gic-v3-its.c new file +64
-0+else + its_data.typer.collid_bits =16 ; + + its_data.typer.virt_lpi = typer &GITS_TYPER_VLPIS ; + its_data.typer.phys_lpi = typer &GITS_TYPER_PLPIS ; +} + +int its_parse_baser (int i,struct its_baser *baser) +{ +void *reg_addr =gicv3_its_base () +GITS_BASER + i *8 ; +u64 val =readq (reg_addr); + +if (!val) { +memset (baser,0 ,sizeof (*baser)); +return -1 ; + } + + baser->valid = val &GITS_BASER_VALID ; + baser->indirect = val &GITS_BASER_INDIRECT ; + baser->type =GITS_BASER_TYPE (val); + baser->esz =GITS_BASER_ENTRY_SIZE (val); + baser->nr_pages =GITS_BASER_NR_PAGES (val); + baser->table_addr = val &GITS_BASER_PHYS_ADDR_MASK ; +switch (val &GITS_BASER_PAGE_SIZE_MASK ) { +case GITS_BASER_PAGE_SIZE_4K : + baser->psz =SZ_4K ; +break ; +case GITS_BASER_PAGE_SIZE_16K : + baser->psz =SZ_16K ; +break ; +case GITS_BASER_PAGE_SIZE_64K : + baser->psz =SZ_64K ; +break ; +default : + baser->psz =SZ_64K ; + } +return 0 ; +} + +struct its_baser *its_lookup_baser (int type) +{ +int i; + +for (i =0 ; i <GITS_BASER_NR_REGS ; i++) { +struct its_baser *baser = &its_data.baser[i]; + +if (baser->type == type) +return baser; + } +return NULL ; +} + +void its_init (void ) +{ +int i; + +if (!its_data.base) +return ; + +its_parse_typer (); +for (i =0 ; i <GITS_BASER_NR_REGS ; i++) +its_parse_baser (i, &its_data.baser[i]); +} +lib/arm/gic.c +6
-3@@ -6,9 +6,11 @@#include <devicetree.h> #include <asm/gic.h> #include <asm/io.h> +#include <asm/gic-v3-its.h> struct gicv2_data gicv2_data;struct gicv3_data gicv3_data; +struct its_data its_data;struct gic_common_ops {void (*enable_defaults)(void );@@ -44,12 +46,13 @@ static const struct gic_common_ops gicv3_common_ops = {* Documentation/devicetree/bindings/interrupt-controller/arm,gic-v3.txt */ static bool -gic_get_dt_bases (const char *compatible,void **base1,void **base2) +gic_get_dt_bases (const char *compatible,void **base1,void **base2,void **base3) {struct dt_pbus_reg reg; -struct dt_device gic; +struct dt_device gic, its;struct dt_bus bus; -int node, ret, i; +int node, subnode, ret, i, len; +const void *fdt =dt_fdt ();dt_bus_init_defaults (&bus);dt_device_init (&gic, &bus,NULL );
lib/arm/gic.c gic_get_dt_bases(const char *compatible, void **base1, void **base2) · line 80
Quoted 4 lines
lib/arm/gic.c +1
-0@@ -74,19 +77,35 @@ gic_get_dt_bases(const char *compatible, void **base1, void **base2) base2[i] =ioremap (reg.addr, reg.size); } +if (base3 && !strcmp (compatible,"arm,gic-v3" )) {
If base != NULL, then we could assert(strcmp(compatible, "arm,cortex-a15-gic") != 0)
Quoted 51 lines
lib/arm/gic.c +18
-2+dt_for_each_subnode (node, subnode) { +const struct fdt_property *prop; + + prop =fdt_get_property (fdt, subnode,"compatible" , &len); +if (!strcmp ((char *)prop->data,"arm,gic-v3-its" )) { +dt_device_bind_node (&its, subnode); + ret =dt_pbus_translate (&its,0 , ®); +assert (ret ==0 ); + *base3 =ioremap (reg.addr, reg.size); +break ; + } + } + + } +return true; }int gicv2_init (void ) {return gic_get_dt_bases ("arm,cortex-a15-gic" , - &gicv2_data.dist_base, &gicv2_data.cpu_base); + &gicv2_data.dist_base, &gicv2_data.cpu_base,NULL ); }int gicv3_init (void ) {return gic_get_dt_bases ("arm,gic-v3" , &gicv3_data.dist_base, - &gicv3_data.redist_bases[0 ]); + &gicv3_data.redist_bases[0 ], &its_data.base); }int gic_version (void )@@ -104,6 +123,7 @@ int gic_init(void) gic_common_ops = &gicv2_common_ops; else if (gicv3_init()) gic_common_ops = &gicv3_common_ops; +its_init ();return gic_version (); }lib/arm64/asm/gic-v3-its.h new file +1
-0@@ -0,0 +1 @@ +#include "../../arm/asm/gic-v3-its.h" Signature · 1 line
2.20.1
Thanks,
drew
Re: [kvm-unit-tests PATCH v3 05/14] arm/arm64: ITS: Introspection tests
Hi Drew,
Quoted 67 lines · Andrew Jones
Quoted 43 lines · Eric Auger
Detect the presence of an ITS as part of the GICv3 init routine, initialize its base address and read few registers the IIDR, the TYPER to store its dimensioning parameters. Also parse the BASER registers.
This is our first ITS test, belonging to a new "its" group.
Signed-off-by Eric Auger <[email protected]>---
v2 -> v3:
- updated dates and changed author
- squash "arm/arm64: ITS: Test BASER" into this patch but removessetup_baserwhich will be introduced later.
- only compile on aarch64
- restrict the new test to aarch64v1 -> v2:
- cleanGITS_TYPERmacros and unused fields in typer struct
- remove memory attribute related macros
- remove everything related to memory attributes
- s/dev_baser/coll_baser/ inreport_info
- add extra line
- removed index filed inits_baser--- arm/Makefile.arm64 | 1 + arm/gic.c | 49 ++++++++++++++++++ arm/unittests.cfg | 7 +++ lib/arm/asm/gic-v3-its.h | 103 +++++++++++++++++++++++++++++++++++++ lib/arm/gic-v3-its.c | 88 +++++++++++++++++++++++++++++++ lib/arm/gic.c | 30 +++++++++-- lib/arm64/asm/gic-v3-its.h | 1 + 7 files changed, 274 insertions(+), 5 deletions(-) create mode 100644 lib/arm/asm/gic-v3-its.h create mode 100644 lib/arm/gic-v3-its.c create mode 100644 lib/arm64/asm/gic-v3-its.harm/Makefile.arm64 +0
-0arm/Makefile.arm64 endef · line 22
Quoted 4 lines
arm/Makefile.arm64 +1
-0@@ -19,6 +19,7 @@ endef cstart.o = $(TEST_DIR)/cstart64.o cflatobjs += lib/arm64/processor.o cflatobjs += lib/arm64/spinlock.o +cflatobjs += lib/arm/gic-v3-its.oIf
gic-v3-its.cwill never be compiled for arm, then it should probably go lib/arm64, not lib/arm. Same comment for all other new source and header files. The only problem with that is...>arm/gic.c line 19
Quoted 12 lines
arm/Makefile.arm64 +0
-0OBJDIRS += lib/arm64arm/gic.c +1
-0@@ -16,6 +16,7 @@#include <asm/processor.h> #include <asm/delay.h> #include <asm/gic.h> +#include <asm/gic-v3-its.h> ...here where we'd have to do
Quoted 6 lines
#if
defined(__aarch64__)
#include <asm/gic-v3-its.h>
#endifwhich is ugly. However that can be avoided by adding an empty
lib/arm/asm/gic-v3-its.hfile.
OK made those changes with this latter containing stubs.
Quoted 9 lines
Quoted 9 lines
arm/gic.c +0
-0#include <asm/smp.h> #include <asm/barrier.h> #include <asm/io.h> @@ -518,6 +519,50 @@ static void gic_test_mmio(void)test_targets (nr_irqs); }
Quoted 8 lines
arm/gic.c static void gic_test_mmio(void) · line 524
Quoted 3 lines
arm/gic.c +3
-0+#if defined(__arm__) + +static void test_its_introspection (void ) {}static void
test_its_introspection(void){ report_abort(...);}
OK
Quoted 87 lines
Quoted 87 lines
arm/gic.c +45
-0+ +#else /* __arch64__ */ + +static void test_its_introspection (void ) +{ +struct its_baser *dev_baser, *coll_baser; +struct its_typer *typer = &its_data.typer; + +if (!gicv3_its_base ()) { +report_skip ("No ITS, skip ..." ); +return ; + } + +/* IIDR */ +report (test_readonly_32 (gicv3_its_base () +GITS_IIDR , false), +"GITS_IIDR is read-only" ), + +/* TYPER */ +report (test_readonly_32 (gicv3_its_base () +GITS_TYPER , false), +"GITS_TYPER is read-only" ); + +report (typer->phys_lpi,"ITS supports physical LPIs" ); +report_info ("vLPI support: %s" , typer->virt_lpi ?"yes" :"no" ); +report_info ("ITT entry size = 0x%x" , typer->ite_size); +report_info ("Bit Count: EventID=%d DeviceId=%d CollId=%d" , + typer->eventid_bits, typer->deviceid_bits, + typer->collid_bits); +report (typer->eventid_bits && typer->deviceid_bits && + typer->collid_bits,"ID spaces" ); +report_info ("Target address format %s" , + typer->pta ?"Redist basse address" :"PE #" ); + + dev_baser =its_lookup_baser (GITS_BASER_TYPE_DEVICE ); + coll_baser =its_lookup_baser (GITS_BASER_TYPE_COLLECTION ); +report (dev_baser && coll_baser,"detect device and collection BASER" ); +report_info ("device baser entry_size = 0x%x" , dev_baser->esz); +report_info ("collection baser entry_size = 0x%x" , coll_baser->esz); +} + +#endif +int main (int argc,char **argv) {if (!gic_init ()) {@@ -549,6 +594,10 @@ int main(int argc, char **argv)report_prefix_push (argv[1 ]);gic_test_mmio ();report_prefix_pop (); + }else if (strcmp (argv[1 ],"its-introspection" ) ==0 ) { +report_prefix_push (argv[1 ]); +test_its_introspection (); +report_prefix_pop (); }else {report_abort ("Unknown subtest '%s'" , argv[1 ]); }arm/unittests.cfg +7
-0@@ -122,6 +122,13 @@ smp = $MAX_SMP extra_params = -machine gic-version=3 -append 'active' groups = gic +[its-introspection] +file = gic.flat +smp = $MAX_SMP +extra_params = -machine gic-version=3 -append 'its-introspection' +groups = its +arch = arm64 + # Test PSCI emulation [psci] file = psci.flatlib/arm/asm/gic-v3-its.h new file +7
-0@@ -0,0 +1,103 @@ +/* +* All ITS* defines are lifted from include/linux/irqchip/arm-gic-v3.h +* +* Copyright (C) 2020, Red Hat Inc, Eric Auger <[email protected]> +* +* This work is licensed under the terms of the GNU LGPL, version 2. +*/
Quoted 8 lines
lib/arm/asm/gic-v3-its.h line 11
Quoted 4 lines
lib/arm/asm/gic-v3-its.h new file +4
-0+#ifndef _ASMARM_GIC_V3_ITS_H_ +#define _ASMARM_GIC_V3_ITS_H_ + +#ifndef __ASSEMBLY__Doesn't look like you use the #else /*
__ASSEMBLY__*/ side of this. I'd leave out the #ifndef until we add defines we need to access from assembler.
OK
Quoted 83 lines
Quoted 83 lines
lib/arm/asm/gic-v3-its.h new file +82
-0+ +struct its_typer { +unsigned int ite_size; +unsigned int eventid_bits; +unsigned int deviceid_bits; +unsigned int collid_bits; +bool pta; +bool phys_lpi; +bool virt_lpi; +}; + +struct its_baser { +int type; +size_t psz; +int nr_pages; +bool indirect; +phys_addr_t table_addr; +bool valid; +int esz; +}; + +#define GITS_BASER_NR_REGS 8 + +struct its_data { +void *base; +struct its_typer typer; +struct its_baser baser[GITS_BASER_NR_REGS ]; +}; + +extern struct its_data its_data; + +#define gicv3_its_base () (its_data.base) + +#if defined(__aarch64__) + +#define GITS_CTLR 0x0000 +#define GITS_IIDR 0x0004 +#define GITS_TYPER 0x0008 +#define GITS_CBASER 0x0080 +#define GITS_CWRITER 0x0088 +#define GITS_CREADR 0x0090 +#define GITS_BASER 0x0100 + +#define GITS_TYPER_PLPIS BIT(0) +#define GITS_TYPER_VLPIS BIT(1) +#define GITS_TYPER_ITT_ENTRY_SIZE GENMASK_ULL(7, 4) +#define GITS_TYPER_ITT_ENTRY_SIZE_SHIFT 4 +#define GITS_TYPER_IDBITS GENMASK_ULL(8, 12) +#define GITS_TYPER_IDBITS_SHIFT 8 +#define GITS_TYPER_DEVBITS GENMASK_ULL(13, 17) +#define GITS_TYPER_DEVBITS_SHIFT 13 +#define GITS_TYPER_PTA BIT(19) +#define GITS_TYPER_CIDBITS GENMASK_ULL(32, 35) +#define GITS_TYPER_CIDBITS_SHIFT 32 +#define GITS_TYPER_CIL BIT(36) + +#define GITS_CTLR_ENABLE (1U << 0) + +#define GITS_CBASER_VALID (1UL << 63) + +#define GITS_BASER_VALID BIT(63) +#define GITS_BASER_INDIRECT BIT(62) +#define GITS_BASER_TYPE_SHIFT (56) +#define GITS_BASER_TYPE (r) (((r) >> GITS_BASER_TYPE_SHIFT) & 7) +#define GITS_BASER_ENTRY_SIZE_SHIFT (48) +#define GITS_BASER_ENTRY_SIZE (r) ((((r) >> GITS_BASER_ENTRY_SIZE_SHIFT) & 0x1f) + 1) +#define GITS_BASER_PAGE_SIZE_SHIFT (8) +#define GITS_BASER_PAGE_SIZE_4K (0UL << GITS_BASER_PAGE_SIZE_SHIFT) +#define GITS_BASER_PAGE_SIZE_16K (1UL << GITS_BASER_PAGE_SIZE_SHIFT) +#define GITS_BASER_PAGE_SIZE_64K (2UL << GITS_BASER_PAGE_SIZE_SHIFT) +#define GITS_BASER_PAGE_SIZE_MASK (3UL << GITS_BASER_PAGE_SIZE_SHIFT) +#define GITS_BASER_PAGES_MAX 256 +#define GITS_BASER_PAGES_SHIFT (0) +#define GITS_BASER_NR_PAGES (r) (((r) & 0xff) + 1) +#define GITS_BASER_PHYS_ADDR_MASK 0xFFFFFFFFF000 +#define GITS_BASER_TYPE_NONE 0 +#define GITS_BASER_TYPE_DEVICE 1 +#define GITS_BASER_TYPE_COLLECTION 4 + +extern void its_parse_typer (void ); +extern void its_init (void ); +extern int its_parse_baser (int i,struct its_baser *baser);
Quoted 8 lines
lib/arm/asm/gic-v3-its.h line 98
Quoted 5 lines
lib/arm/asm/gic-v3-its.h new file +5
-0+extern struct its_baser *its_lookup_baser (int type); + +#else /* __arm__ */ + +static inline void its_init (void ) {}Looks like the empty
gic-v3-its.hI suggested creating above will actually be useful. We can add stubs like this in it.
OK
Quoted 42 lines
Quoted 26 lines
lib/arm/asm/gic-v3-its.h new file +5
-0+ +#endif + +#endif /* !__ASSEMBLY__ */ +#endif /* _ASMARM_GIC_V3_ITS_H_ */ lib/arm/gic-v3-its.c new file +14
-0@@ -0,0 +1,88 @@ +/* +* Copyright (C) 2020, Red Hat Inc, Eric Auger <[email protected]> +* +* This work is licensed under the terms of the GNU LGPL, version 2. +*/ +#include <asm/gic.h> +#include <alloc_page.h> +#include <asm/gic-v3-its.h> + +void its_parse_typer (void ) +{ +u64 typer =readq (gicv3_its_base () +GITS_TYPER ); + + its_data.typer.ite_size = ((typer & GITS_TYPER_ITT_ENTRY_SIZE) >>lib/arm/gic-v3-its.c line 24
Quoted 10 lines
lib/arm/gic-v3-its.c new file +10
-0+ GITS_TYPER_ITT_ENTRY_SIZE_SHIFT) + 1; + its_data.typer.pta = typer &GITS_TYPER_PTA ; + its_data.typer.eventid_bits = ((typer &GITS_TYPER_IDBITS ) >> +GITS_TYPER_IDBITS_SHIFT ) +1 ; + its_data.typer.deviceid_bits = ((typer &GITS_TYPER_DEVBITS ) >> +GITS_TYPER_DEVBITS_SHIFT ) +1 ; + +if (typer &GITS_TYPER_CIL ) + its_data.typer.collid_bits = ((typer &GITS_TYPER_CIDBITS ) >> +GITS_TYPER_CIDBITS_SHIFT ) +1 ;nit: please consider aligning like this
((typer & MASK) >>
SHIFT) + 1;
Quoted 8 lines
Or, maybe better to macro it
#define
TYPER_FIELD(typer, mask, shift)(((type) & (mask) >> (shift)) + 1)And, rather than have a bunch of 'its_data.typer's we could use an alias, helping us stay within 120 chars.
struct
its_typer*t = &its_data.typer;
yes used that trick and relied on 120 char max.
Quoted 106 lines
t->ite_size= TYPER_FIELD(typer,GITS_TYPER_ITT_ENTRY_SIZE,GITS_TYPER_ITT_ENTRY_SIZE_SHIFT);Quoted 101 lines
lib/arm/gic-v3-its.c new file +64
-0+else + its_data.typer.collid_bits =16 ; + + its_data.typer.virt_lpi = typer &GITS_TYPER_VLPIS ; + its_data.typer.phys_lpi = typer &GITS_TYPER_PLPIS ; +} + +int its_parse_baser (int i,struct its_baser *baser) +{ +void *reg_addr =gicv3_its_base () +GITS_BASER + i *8 ; +u64 val =readq (reg_addr); + +if (!val) { +memset (baser,0 ,sizeof (*baser)); +return -1 ; + } + + baser->valid = val &GITS_BASER_VALID ; + baser->indirect = val &GITS_BASER_INDIRECT ; + baser->type =GITS_BASER_TYPE (val); + baser->esz =GITS_BASER_ENTRY_SIZE (val); + baser->nr_pages =GITS_BASER_NR_PAGES (val); + baser->table_addr = val &GITS_BASER_PHYS_ADDR_MASK ; +switch (val &GITS_BASER_PAGE_SIZE_MASK ) { +case GITS_BASER_PAGE_SIZE_4K : + baser->psz =SZ_4K ; +break ; +case GITS_BASER_PAGE_SIZE_16K : + baser->psz =SZ_16K ; +break ; +case GITS_BASER_PAGE_SIZE_64K : + baser->psz =SZ_64K ; +break ; +default : + baser->psz =SZ_64K ; + } +return 0 ; +} + +struct its_baser *its_lookup_baser (int type) +{ +int i; + +for (i =0 ; i <GITS_BASER_NR_REGS ; i++) { +struct its_baser *baser = &its_data.baser[i]; + +if (baser->type == type) +return baser; + } +return NULL ; +} + +void its_init (void ) +{ +int i; + +if (!its_data.base) +return ; + +its_parse_typer (); +for (i =0 ; i <GITS_BASER_NR_REGS ; i++) +its_parse_baser (i, &its_data.baser[i]); +} +lib/arm/gic.c +6
-3@@ -6,9 +6,11 @@#include <devicetree.h> #include <asm/gic.h> #include <asm/io.h> +#include <asm/gic-v3-its.h> struct gicv2_data gicv2_data;struct gicv3_data gicv3_data; +struct its_data its_data;struct gic_common_ops {void (*enable_defaults)(void );@@ -44,12 +46,13 @@ static const struct gic_common_ops gicv3_common_ops = {* Documentation/devicetree/bindings/interrupt-controller/arm,gic-v3.txt */ static bool -gic_get_dt_bases (const char *compatible,void **base1,void **base2) +gic_get_dt_bases (const char *compatible,void **base1,void **base2,void **base3) {struct dt_pbus_reg reg; -struct dt_device gic; +struct dt_device gic, its;struct dt_bus bus; -int node, ret, i; +int node, subnode, ret, i, len; +const void *fdt =dt_fdt ();dt_bus_init_defaults (&bus);dt_device_init (&gic, &bus,NULL );@@ -74,19 +77,35 @@ gic_get_dt_bases(const char *compatible, void **base1, void **base2) base2[i] =ioremap (reg.addr, reg.size); }
Quoted 3 lines
lib/arm/gic.c gic_get_dt_bases(const char *compatible, void **base1, void **base2) · line 80
Quoted 1 line
lib/arm/gic.c +1
-0+if (base3 && !strcmp (compatible,"arm,gic-v3" )) {If base != NULL, then we could
assert(strcmp(compatible, "arm,cortex-a15-gic") != 0)
OK
Thanks
Eric
Quoted 56 lines
Quoted 52 lines
lib/arm/gic.c +18
-2+dt_for_each_subnode (node, subnode) { +const struct fdt_property *prop; + + prop =fdt_get_property (fdt, subnode,"compatible" , &len); +if (!strcmp ((char *)prop->data,"arm,gic-v3-its" )) { +dt_device_bind_node (&its, subnode); + ret =dt_pbus_translate (&its,0 , ®); +assert (ret ==0 ); + *base3 =ioremap (reg.addr, reg.size); +break ; + } + } + + } +return true; }int gicv2_init (void ) {return gic_get_dt_bases ("arm,cortex-a15-gic" , - &gicv2_data.dist_base, &gicv2_data.cpu_base); + &gicv2_data.dist_base, &gicv2_data.cpu_base,NULL ); }int gicv3_init (void ) {return gic_get_dt_bases ("arm,gic-v3" , &gicv3_data.dist_base, - &gicv3_data.redist_bases[0 ]); + &gicv3_data.redist_bases[0 ], &its_data.base); }int gic_version (void )@@ -104,6 +123,7 @@ int gic_init(void) gic_common_ops = &gicv2_common_ops; else if (gicv3_init()) gic_common_ops = &gicv3_common_ops; +its_init ();return gic_version (); }lib/arm64/asm/gic-v3-its.h new file +1
-0@@ -0,0 +1 @@ +#include "../../arm/asm/gic-v3-its.h" Signature · 1 line
2.20.1Thanks,
drew
PATCH v3 6/14 arm/arm64: gicv3: Set the LPI config and pending tables
In reply to Eric AugerAllocate the LPI configuration and per re-distributor pending table. Set redistributor's PROPBASER and PENDBASER. The LPIs are enabled by default in the config table.
Also introduce a helper routine that allows to set the pending table bit for a given LPI.
---
v2 -> v3:
- Move the helpers in lib/arm/gic-v3.c and prefix them with "gicv3_" and add _lpi prefix too
v1 -> v2:
- remove memory attributes
--- lib/arm/asm/gic-v3.h | 16 +++++++++++ lib/arm/gic-v3.c | 64 ++++++++++++++++++++++++++++++++++++++++++++ 2 files changed, 80 insertions(+)
lib/arm/asm/gic-v3.h +16 -0
@@ -48,6 +48,16 @@#define MPIDR_TO_SGI_AFFINITY (cluster_id, level) \ (MPIDR_AFFINITY_LEVEL(cluster_id, level) << ICC_SGI1R_AFFINITY_## level ## _SHIFT) +#define GICR_PROPBASER_IDBITS_MASK (0x1f) + +#define GICR_PENDBASER_PTZ BIT_ULL(62) + +#define LPI_PROP_GROUP1 (1 << 1) +#define LPI_PROP_ENABLED (1 << 0) +#define LPI_PROP_DEFAULT_PRIO 0xa0 +#define LPI_PROP_DEFAULT (LPI_PROP_DEFAULT_PRIO | LPI_PROP_GROUP1 | \ + LPI_PROP_ENABLED) +#include <asm/arch_gicv3.h> #ifndef __ASSEMBLY__
@@ -64,6 +74,8 @@ struct gicv3_data {void *dist_base;void *redist_bases[GICV3_NR_REDISTS ];void *redist_base[NR_CPUS ]; +void *lpi_prop; +void *lpi_pend[NR_CPUS ];unsigned int irq_nr; };extern struct gicv3_data gicv3_data;
@@ -80,6 +92,10 @@ extern void gicv3_write_eoir(u32 irqstat);extern void gicv3_ipi_send_single (int irq,int cpu);extern void gicv3_ipi_send_mask (int irq,const cpumask_t *dest);extern void gicv3_set_redist_base (size_t stride); +extern void gicv3_lpi_set_config (int n,u8 val); +extern u8 gicv3_lpi_get_config (int n); +extern void gicv3_lpi_set_pending_table_bit (int rdist,int n,bool set); +extern void gicv3_lpi_alloc_tables (void );static inline void gicv3_do_wait_for_rwp (void *base) {
lib/arm/gic-v3.c +64 -0
73 lines not on this page · raw message
Signature · 1 line
2.20.1
Zenghui Yu
reviewerRe: [kvm-unit-tests PATCH v3 06/14] arm/arm64: gicv3: Set the LPI config and pending tables
Hi Eric,
Quoted 27 lines · Eric Auger
Allocate the LPI configuration and per re-distributor pending table. Set redistributor's PROPBASER and PENDBASER. The LPIs are enabled by default in the config table.
Also introduce a helper routine that allows to set the pending table bit for a given LPI.
Signed-off-by Eric Auger <[email protected]>---
v2 -> v3:
- Move the helpers inlib/arm/gic-v3.cand prefix them with "gicv3_" and add_lpiprefix toov1 -> v2:
- remove memory attributes--- lib/arm/asm/gic-v3.h | 16 +++++++++++ lib/arm/gic-v3.c | 64 ++++++++++++++++++++++++++++++++++++++++++++ 2 files changed, 80 insertions(+)lib/arm/asm/gic-v3.h +0
-0
lib/arm/asm/gic-v3.h line 51
Quoted 4 lines
lib/arm/asm/gic-v3.h +1
-0@@ -48,6 +48,16 @@#define MPIDR_TO_SGI_AFFINITY (cluster_id, level) \ (MPIDR_AFFINITY_LEVEL(cluster_id, level) << ICC_SGI1R_AFFINITY_## level ## _SHIFT) +#define GICR_PROPBASER_IDBITS_MASK (0x1f)
This is not being used. You can use it when calculating prop_val or just drop it.
Quoted 76 lines
lib/arm/asm/gic-v3.h +15
-0+ +#define GICR_PENDBASER_PTZ BIT_ULL(62) + +#define LPI_PROP_GROUP1 (1 << 1) +#define LPI_PROP_ENABLED (1 << 0) +#define LPI_PROP_DEFAULT_PRIO 0xa0 +#define LPI_PROP_DEFAULT (LPI_PROP_DEFAULT_PRIO | LPI_PROP_GROUP1 | \ + LPI_PROP_ENABLED) +#include <asm/arch_gicv3.h> #ifndef __ASSEMBLY__@@ -64,6 +74,8 @@ struct gicv3_data {void *dist_base;void *redist_bases[GICV3_NR_REDISTS ];void *redist_base[NR_CPUS ]; +void *lpi_prop; +void *lpi_pend[NR_CPUS ];unsigned int irq_nr; };extern struct gicv3_data gicv3_data;@@ -80,6 +92,10 @@ extern void gicv3_write_eoir(u32 irqstat);extern void gicv3_ipi_send_single (int irq,int cpu);extern void gicv3_ipi_send_mask (int irq,const cpumask_t *dest);extern void gicv3_set_redist_base (size_t stride); +extern void gicv3_lpi_set_config (int n,u8 val); +extern u8 gicv3_lpi_get_config (int n); +extern void gicv3_lpi_set_pending_table_bit (int rdist,int n,bool set); +extern void gicv3_lpi_alloc_tables (void );static inline void gicv3_do_wait_for_rwp (void *base) {lib/arm/gic-v3.c +29
-0@@ -5,6 +5,7 @@*/ #include <asm/gic.h> #include <asm/io.h> +#include <alloc_page.h> void gicv3_set_redist_base (size_t stride) {@@ -147,3 +148,66 @@ void gicv3_ipi_send_single(int irq, int cpu)cpumask_set_cpu (cpu, &dest);gicv3_ipi_send_mask (irq, &dest); } + +#if defined(__aarch64__) +/* alloc_lpi_tables: Allocate LPI config and pending tables */ +void gicv3_lpi_alloc_tables (void ) +{ +unsigned long n =SZ_64K >>PAGE_SHIFT ; +unsigned long order =fls (n); +u64 prop_val; +int cpu; + + gicv3_data.lpi_prop = (void *)virt_to_phys (alloc_pages (order)); + +/* ID bits = 13, ie. up to 14b LPI INTID */ + prop_val = (u64 )gicv3_data.lpi_prop |13 ; + +/* +* Allocate pending tables for each redistributor +* and set PROPBASER and PENDBASER +*/ +for_each_present_cpu (cpu) { +u64 pend_val; +void *ptr; + + ptr = gicv3_data.redist_base[cpu]; + +writeq (prop_val, ptr +GICR_PROPBASER ); + + gicv3_data.lpi_pend[cpu] = (void *)virt_to_phys (alloc_pages (order));
lib/arm/gic-v3.c void gicv3_ipi_send_single(int irq, int cpu) · line 188
Quoted 10 lines
lib/arm/gic-v3.c +10
-0+ + pend_val = (u64 )gicv3_data.lpi_pend[cpu]; + +writeq (pend_val, ptr +GICR_PENDBASER ); + } +} + +void gicv3_lpi_set_config (int n,u8 value) +{ +u8 *entry = (u8 *)(gicv3_data.lpi_prop + (n -8192 ));
But this is actually the *physical* address, shouldn't it be converted by phys_to_virt() before reading/writing something? Like what you've done for the 'lpi_pend[rdist]' before writing pending bit. Or I'm missing some points here?
lib/arm/gic-v3.c void gicv3_ipi_send_single(int irq, int cpu) · line 195
Quoted 7 lines
lib/arm/gic-v3.c +7
-0+ + *entry = value; +} + +u8 gicv3_lpi_get_config (int n) +{ +u8 *entry = (u8 *)(gicv3_data.lpi_prop + (n -8192 ));
The same as above.
Thanks,
Zenghui
Quoted 19 lines
lib/arm/gic-v3.c +18
-0+ + return *entry; +} + +void gicv3_lpi_set_pending_table_bit (int rdist,int n,bool set) +{ +u8 *ptr =phys_to_virt ((phys_addr_t )gicv3_data.lpi_pend[rdist]); +u8 mask =1 << (n %8 ), byte; + + ptr += (n /8 ); + byte = *ptr; +if (set) + byte |= mask; +else + byte &= ~mask; + *ptr = byte; +} +#endif /* __aarch64__ */
Re: [kvm-unit-tests PATCH v3 06/14] arm/arm64: gicv3: Set the LPI config and pending tables
Hi Zenghui,
Quoted 25 lines · Zenghui Yu
Hi Eric,
Quoted 22 lines · Eric Auger
Allocate the LPI configuration and per re-distributor pending table. Set redistributor's PROPBASER and PENDBASER. The LPIs are enabled by default in the config table.
Also introduce a helper routine that allows to set the pending table bit for a given LPI.
Signed-off-by Eric Auger <[email protected]>---
v2 -> v3:
- Move the helpers inlib/arm/gic-v3.cand prefix them with "gicv3_" and add_lpiprefix toov1 -> v2:
- remove memory attributes
---
lib/arm/asm/gic-v3.h| 16 +++++++++++
lib/arm/gic-v3.c | 64 ++++++++++++++++++++++++++++++++++++++++++++2 files changed, 80 insertions(+)
Quoted 12 lines
Quoted 9 lines
lib/arm/asm/gic-v3.h +0
-0@@ -48,6 +48,16 @@ #define MPIDR_TO_SGI_AFFINITY(cluster_id, level) \(
MPIDR_AFFINITY_LEVEL(cluster_id, level)<< ICC_SGI1R_AFFINITY_## level ##_SHIFT)
+#define GICR_PROPBASER_IDBITS_MASK (0x1f)This is not being used. You can use it when calculating
prop_valor just drop it.
yep dropped it.
Quoted 82 lines
Quoted 82 lines
lib/arm/asm/gic-v3.h +9
-0+ +#define GICR_PENDBASER_PTZ BIT_ULL(62) + +#define LPI_PROP_GROUP1 (1 << 1) +#define LPI_PROP_ENABLED (1 << 0) +#define LPI_PROP_DEFAULT_PRIO 0xa0 +#define LPI_PROP_DEFAULT (LPI_PROP_DEFAULT_PRIO | LPI_PROP_GROUP1 | \ + LPI_PROP_ENABLED) +#include <
asm/arch_gicv3.h>
#ifndef__ASSEMBLY__lib/arm/asm/gic-v3.h +6
-0@@ -64,6 +74,8 @@ struct gicv3_data { void *dist_base; void *redist_bases[GICV3_NR_REDISTS]; void *redist_base[NR_CPUS]; + void *lpi_prop; + void *lpi_pend[NR_CPUS]; unsigned int irq_nr; }; extern struct gicv3_data gicv3_data;@@ -80,6 +92,10 @@ extern void gicv3_write_eoir(u32 irqstat); extern void gicv3_ipi_send_single(int irq, int cpu); extern void gicv3_ipi_send_mask(int irq, const cpumask_t *dest); extern void gicv3_set_redist_base(size_t stride); +extern void gicv3_lpi_set_config (int n,u8 val); +extern u8 gicv3_lpi_get_config (int n); +extern void gicv3_lpi_set_pending_table_bit (int rdist,int n,bool set); +extern void gicv3_lpi_alloc_tables (void ); static inline void gicv3_do_wait_for_rwp(void *base) {lib/arm/gic-v3.c +29
-0@@ -5,6 +5,7 @@*/ #include <asm/gic.h> #include <asm/io.h> +#include <alloc_page.h> void gicv3_set_redist_base(size_t stride) {@@ -147,3 +148,66 @@ void gicv3_ipi_send_single(int irq, int cpu) cpumask_set_cpu(cpu, &dest); gicv3_ipi_send_mask(irq, &dest); } + +#if defined(__aarch64__) +/* alloc_lpi_tables: Allocate LPI config and pending tables */ +void gicv3_lpi_alloc_tables (void ) +{ + unsigned long n = SZ_64K >> PAGE_SHIFT; + unsigned long order = fls(n); + u64 prop_val; + int cpu; + + gicv3_data.lpi_prop = (void *)virt_to_phys(alloc_pages(order)); + + /* ID bits = 13, ie. up to 14b LPI INTID */ + prop_val = (u64)gicv3_data.lpi_prop | 13; + + /* + * Allocate pending tables for each redistributor + * and set PROPBASER and PENDBASER + */ + for_each_present_cpu(cpu) { + u64 pend_val; + void *ptr; + + ptr = gicv3_data.redist_base[cpu]; + + writeq(prop_val, ptr + GICR_PROPBASER); + + gicv3_data.lpi_pend[cpu] = (void*)virt_to_phys(alloc_pages(order));
lib/arm/gic-v3.c +7
-0+ + pend_val = (u64)gicv3_data.lpi_pend[cpu]; + + writeq(pend_val, ptr + GICR_PENDBASER); + } +} +
Quoted 8 lines
lib/arm/gic-v3.c void gicv3_ipi_send_single(int irq, int cpu) · line 188
Quoted 3 lines
lib/arm/gic-v3.c +3
-0+void gicv3_lpi_set_config (int n,u8 value) +{ + u8 *entry = (u8 *)(gicv3_data.lpi_prop + (n - 8192));But this is actually the *physical* address, shouldn't it be converted by
phys_to_virt()before reading/writing something? Like what you've done for the 'lpi_pend[rdist]' before writing pending bit. Or I'm missing some points here?
Agreed! Thanks
Eric
Quoted 35 lines
lib/arm/gic-v3.c void gicv3_ipi_send_single(int irq, int cpu) · line 195
Quoted 8 lines
lib/arm/gic-v3.c +7
-0+ + *entry = value; +} + +u8 gicv3_lpi_get_config (int n) +{ + u8 *entry = (u8 *)(gicv3_data.lpi_prop + (n - 8192));The same as above.
Thanks,
ZenghuiQuoted 19 lines
lib/arm/gic-v3.c +18
-0+ + return *entry; +} + +void gicv3_lpi_set_pending_table_bit (int rdist,int n,bool set) +{ + u8 *ptr = phys_to_virt((phys_addr_t)gicv3_data.lpi_pend[rdist]); + u8 mask = 1 << (n % 8), byte; + + ptr += (n / 8); + byte = *ptr; + if (set) + byte |= mask; + else + byte &= ~mask; + *ptr = byte; +} +#endif /* __aarch64__ */
Re: [kvm-unit-tests PATCH v3 06/14] arm/arm64: gicv3: Set the LPI config and pending tables
In reply to Eric AugerQuoted 27 lines · Eric Auger
Allocate the LPI configuration and per re-distributor pending table. Set redistributor's PROPBASER and PENDBASER. The LPIs are enabled by default in the config table.
Also introduce a helper routine that allows to set the pending table bit for a given LPI.
Signed-off-by Eric Auger <[email protected]>---
v2 -> v3:
- Move the helpers inlib/arm/gic-v3.cand prefix them with "gicv3_" and add_lpiprefix toov1 -> v2:
- remove memory attributes--- lib/arm/asm/gic-v3.h | 16 +++++++++++ lib/arm/gic-v3.c | 64 ++++++++++++++++++++++++++++++++++++++++++++ 2 files changed, 80 insertions(+)lib/arm/asm/gic-v3.h +0
-0
lib/arm/asm/gic-v3.h line 53
Quoted 6 lines
lib/arm/asm/gic-v3.h +3
-0@@ -48,6 +48,16 @@#define MPIDR_TO_SGI_AFFINITY (cluster_id, level) \ (MPIDR_AFFINITY_LEVEL(cluster_id, level) << ICC_SGI1R_AFFINITY_## level ## _SHIFT) +#define GICR_PROPBASER_IDBITS_MASK (0x1f) + +#define GICR_PENDBASER_PTZ BIT_ULL(62)
Strange indent here, too far and not tabs
Quoted 68 lines
lib/arm/asm/gic-v3.h +13
-0+ +#define LPI_PROP_GROUP1 (1 << 1) +#define LPI_PROP_ENABLED (1 << 0) +#define LPI_PROP_DEFAULT_PRIO 0xa0 +#define LPI_PROP_DEFAULT (LPI_PROP_DEFAULT_PRIO | LPI_PROP_GROUP1 | \ + LPI_PROP_ENABLED) +#include <asm/arch_gicv3.h> #ifndef __ASSEMBLY__@@ -64,6 +74,8 @@ struct gicv3_data {void *dist_base;void *redist_bases[GICV3_NR_REDISTS ];void *redist_base[NR_CPUS ]; +void *lpi_prop; +void *lpi_pend[NR_CPUS ];unsigned int irq_nr; };extern struct gicv3_data gicv3_data;@@ -80,6 +92,10 @@ extern void gicv3_write_eoir(u32 irqstat);extern void gicv3_ipi_send_single (int irq,int cpu);extern void gicv3_ipi_send_mask (int irq,const cpumask_t *dest);extern void gicv3_set_redist_base (size_t stride); +extern void gicv3_lpi_set_config (int n,u8 val); +extern u8 gicv3_lpi_get_config (int n); +extern void gicv3_lpi_set_pending_table_bit (int rdist,int n,bool set); +extern void gicv3_lpi_alloc_tables (void );static inline void gicv3_do_wait_for_rwp (void *base) {lib/arm/gic-v3.c +23
-0@@ -5,6 +5,7 @@*/ #include <asm/gic.h> #include <asm/io.h> +#include <alloc_page.h> void gicv3_set_redist_base (size_t stride) {@@ -147,3 +148,66 @@ void gicv3_ipi_send_single(int irq, int cpu)cpumask_set_cpu (cpu, &dest);gicv3_ipi_send_mask (irq, &dest); } + +#if defined(__aarch64__) +/* alloc_lpi_tables: Allocate LPI config and pending tables */ +void gicv3_lpi_alloc_tables (void ) +{ +unsigned long n =SZ_64K >>PAGE_SHIFT ; +unsigned long order =fls (n); +u64 prop_val; +int cpu; + + gicv3_data.lpi_prop = (void *)virt_to_phys (alloc_pages (order)); + +/* ID bits = 13, ie. up to 14b LPI INTID */ + prop_val = (u64 )gicv3_data.lpi_prop |13 ; + +/* +* Allocate pending tables for each redistributor +* and set PROPBASER and PENDBASER +*/ +for_each_present_cpu (cpu) { +u64 pend_val; +void *ptr;
lib/arm/gic-v3.c void gicv3_ipi_send_single(int irq, int cpu) · line 182
Quoted 10 lines
lib/arm/gic-v3.c +10
-0+ + ptr = gicv3_data.redist_base[cpu]; + +writeq (prop_val, ptr +GICR_PROPBASER ); + + gicv3_data.lpi_pend[cpu] = (void *)virt_to_phys (alloc_pages (order)); + + pend_val = (u64 )gicv3_data.lpi_pend[cpu]; + +writeq (pend_val, ptr +GICR_PENDBASER );
nit: probably don't need all the blank lines
Quoted 6 lines
lib/arm/gic-v3.c +6
-0+ } +} + +void gicv3_lpi_set_config (int n,u8 value) +{ +u8 *entry = (u8 *)(gicv3_data.lpi_prop + (n -8192 ));
lib/arm/gic-v3.c void gicv3_ipi_send_single(int irq, int cpu) · line 198
Quoted 10 lines
lib/arm/gic-v3.c +10
-0+ + *entry = value; +} + +u8 gicv3_lpi_get_config (int n) +{ +u8 *entry = (u8 *)(gicv3_data.lpi_prop + (n -8192 )); + +return *entry; +}
We probably need a define for the 8192. LPI_ID_BASE? And, like PPI and SPI maybe we need an LPI(id)? '#define LPI(id) ((id) + LPI_ID_BASE)'?
Also, is lpi_prop an array of u8's? If so, then I'd think we could declare it as a u8 * in gicv3_data. Then, instead of the above two functions just do something like
#define gicv3_lpi_set_config(n, val) ({
gicv3_data.lpi_prop[n] = (val); })
#define gicv3_lpi_get_config(n) (gicv3_data.lpi_prop[n])
lib/arm/gic-v3.c void gicv3_ipi_send_single(int irq, int cpu) · line 200
Quoted 2 lines
lib/arm/gic-v3.c +2
-0+ +void gicv3_lpi_set_pending_table_bit (int rdist,int n,bool set)
These types of functions that can be used both as a setter and clearer usually have a name like foo_set_clr_bar(). That would make this long name a bit longer, but do we need the '_table_bit' part of the name?
lib/arm/gic-v3.c void gicv3_ipi_send_single(int irq, int cpu) · line 202
Quoted 2 lines
lib/arm/gic-v3.c +2
-0+{ +u8 *ptr =phys_to_virt ((phys_addr_t )gicv3_data.lpi_pend[rdist]);
If lpi_pend is a physical address then why not declare it like that in gicv3_data, i.e. phys_addr_t lpi_pend[NR_CPUS] ?
lib/arm/gic-v3.c void gicv3_ipi_send_single(int irq, int cpu) · line 206
Quoted 4 lines
lib/arm/gic-v3.c +4
-0+u8 mask =1 << (n %8 ), byte; + + ptr += (n /8 ); + byte = *ptr;
And Zenghui already pointed out that this doesn't look right for a physical address.
Quoted 11 lines
lib/arm/gic-v3.c +7
-0+if (set) + byte |= mask; +else + byte &= ~mask; + *ptr = byte; +} +#endif /* __aarch64__ */ Signature · 1 line
2.20.1
Thanks,
drew
PATCH v3 7/14 arm/arm64: gicv3: Enable/Disable LPIs at re-distributor level
In reply to Eric AugerThis helper function controls the signaling of LPIs at redistributor level.
---
v2 -> v3:
- move the helper in lib/arm/gic-v3.c
- rename the function with gicv3_lpi_ prefix
- s/report_abort/assert
--- lib/arm/asm/gic-v3.h | 1 + lib/arm/gic-v3.c | 17 +++++++++++++++++ 2 files changed, 18 insertions(+)
lib/arm/asm/gic-v3.h +1 -0
@@ -96,6 +96,7 @@ extern void gicv3_lpi_set_config(int n, u8 val);extern u8 gicv3_lpi_get_config (int n);extern void gicv3_lpi_set_pending_table_bit (int rdist,int n,bool set);extern void gicv3_lpi_alloc_tables (void ); +extern void gicv3_lpi_rdist_ctrl (u32 redist,bool set);static inline void gicv3_do_wait_for_rwp (void *base) {
lib/arm/gic-v3.c +17 -0
@@ -210,4 +210,21 @@ void gicv3_lpi_set_pending_table_bit(int rdist, int n, bool set) byte &= ~mask; *ptr = byte; } + +void gicv3_lpi_rdist_ctrl (u32 redist,bool set) +{ +void *ptr; +u64 val; + +assert (redist < nr_cpus); + + ptr = gicv3_data.redist_base[redist]; + val =readl (ptr +GICR_CTLR ); +if (set) + val |=GICR_CTLR_ENABLE_LPIS ; +else + val &= ~GICR_CTLR_ENABLE_LPIS ; +writel (val, ptr +GICR_CTLR ); +}#endif /* __aarch64__ */ +
Signature · 1 line
2.20.1
Re: [kvm-unit-tests PATCH v3 07/14] arm/arm64: gicv3: Enable/Disable LPIs at re-distributor level
Quoted 33 lines · Eric Auger
This helper function controls the signaling of LPIs at redistributor level.
Signed-off-by Eric Auger <[email protected]>---
v2 -> v3:
- move the helper inlib/arm/gic-v3.c
- rename the function withgicv3_lpi_prefix
- s/report_abort/assert--- lib/arm/asm/gic-v3.h | 1 + lib/arm/gic-v3.c | 17 +++++++++++++++++ 2 files changed, 18 insertions(+)lib/arm/asm/gic-v3.h +1
-0@@ -96,6 +96,7 @@ extern void gicv3_lpi_set_config(int n, u8 val);extern u8 gicv3_lpi_get_config (int n);extern void gicv3_lpi_set_pending_table_bit (int rdist,int n,bool set);extern void gicv3_lpi_alloc_tables (void ); +extern void gicv3_lpi_rdist_ctrl (u32 redist,bool set);static inline void gicv3_do_wait_for_rwp (void *base) {lib/arm/gic-v3.c +0
-0
lib/arm/gic-v3.c void gicv3_lpi_set_pending_table_bit(int rdist, int n, bool set) · line 214
Quoted 5 lines
lib/arm/gic-v3.c +2
-0@@ -210,4 +210,21 @@ void gicv3_lpi_set_pending_table_bit(int rdist, int n, bool set) byte &= ~mask; *ptr = byte; } + +void gicv3_lpi_rdist_ctrl (u32 redist,bool set)
_set_clr_ ?
Quoted 6 lines
lib/arm/gic-v3.c +6
-0+{ +void *ptr; +u64 val; + +assert (redist < nr_cpus); +
lib/arm/gic-v3.c void gicv3_lpi_set_pending_table_bit(int rdist, int n, bool set) · line 230
Quoted 10 lines
lib/arm/gic-v3.c +9
-0+ ptr = gicv3_data.redist_base[redist]; + val =readl (ptr +GICR_CTLR ); +if (set) + val |=GICR_CTLR_ENABLE_LPIS ; +else + val &= ~GICR_CTLR_ENABLE_LPIS ; + writel(val, ptr + GICR_CTLR); +}#endif /* __aarch64__ */ +
stray blank line here
Quoted 3 lines
Signature · 1 line
2.20.1
I'm not sure why this needs its own patch. I could just be part of the next patch.
Thanks,
drew
Re: [kvm-unit-tests PATCH v3 07/14] arm/arm64: gicv3: Enable/Disable LPIs at re-distributor level
Quoted 27 lines · Andrew Jones
Quoted 26 lines · Eric Auger
This helper function controls the signaling of LPIs at redistributor level.
Signed-off-by Eric Auger <[email protected]>---
v2 -> v3:
- move the helper inlib/arm/gic-v3.c
- rename the function withgicv3_lpi_prefix
- s/report_abort/assert--- lib/arm/asm/gic-v3.h | 1 + lib/arm/gic-v3.c | 17 +++++++++++++++++ 2 files changed, 18 insertions(+)lib/arm/asm/gic-v3.h +1
-0@@ -96,6 +96,7 @@ extern void gicv3_lpi_set_config(int n, u8 val);extern u8 gicv3_lpi_get_config (int n);extern void gicv3_lpi_set_pending_table_bit (int rdist,int n,bool set);extern void gicv3_lpi_alloc_tables (void ); +extern void gicv3_lpi_rdist_ctrl (u32 redist,bool set);
Quoted 14 lines
lib/arm/gic-v3.c void gicv3_lpi_set_pending_table_bit(int rdist, int n, bool set) · line 214
Quoted 12 lines
lib/arm/asm/gic-v3.h +0
-0static inline void gicv3_do_wait_for_rwp (void *base) {lib/arm/gic-v3.c +2
-0@@ -210,4 +210,21 @@ void gicv3_lpi_set_pending_table_bit(int rdist, int n, bool set) byte &= ~mask; *ptr = byte; } + +void gicv3_lpi_rdist_ctrl (u32 redist,bool set)_set_clr_ ?
No, probably not _set_clr_ here. The function could be static though, with other functions to enable/disable
void gicv3_lpi_rdist_enable(redist) { gicv3_lpi_rdist_ctrl(redist, true); } void gicv3_lpi_rdist_disable(redist) { gicv3_lpi_rdist_ctrl(redist, false); }
But whatever.
Quoted 30 lines
Quoted 7 lines
lib/arm/gic-v3.c +6
-0+{ +void *ptr; +u64 val; + +assert (redist < nr_cpus); +lib/arm/gic-v3.c void gicv3_lpi_set_pending_table_bit(int rdist, int n, bool set) · line 230
Quoted 10 lines
lib/arm/gic-v3.c +9
-0+ ptr = gicv3_data.redist_base[redist]; + val =readl (ptr +GICR_CTLR ); +if (set) + val |=GICR_CTLR_ENABLE_LPIS ; +else + val &= ~GICR_CTLR_ENABLE_LPIS ; + writel(val, ptr + GICR_CTLR); +}#endif /* __aarch64__ */ +stray blank line here
Quoted 3 lines
Signature · 1 line
2.20.1I'm not sure why this needs its own patch. I could just be part of the next patch.
Thanks,
drew
PATCH v3 8/14 arm/arm64: ITS: its_enable_defaults
In reply to Eric Augerits_enable_defaults() is the top init function that allocates the command queue and all the requested tables (device, collection, lpi config and pending tables), enable LPIs at distributor level and ITS level.
gicv3_enable_defaults must be called before.
---
v2 -> v3:
- introduce its_setup_baser in this patch
- squash "arm/arm64: ITS: Init the command queue" in this patch.
--- lib/arm/asm/gic-v3-its.h | 8 ++++ lib/arm/gic-v3-its.c | 89 ++++++++++++++++++++++++++++++++++++++++ 2 files changed, 97 insertions(+)
lib/arm/asm/gic-v3-its.h +8 -0
@@ -36,6 +36,8 @@ struct its_data {void *base;struct its_typer typer;struct its_baser baser[GITS_BASER_NR_REGS ]; +struct its_cmd_block *cmd_base; +struct its_cmd_block *cmd_write; };extern struct its_data its_data;
@@ -88,10 +90,16 @@ extern struct its_data its_data;#define GITS_BASER_TYPE_DEVICE 1#define GITS_BASER_TYPE_COLLECTION 4 + +struct its_cmd_block { +u64 raw_cmd[4 ]; +}; +extern void its_parse_typer (void );extern void its_init (void );extern int its_parse_baser (int i,struct its_baser *baser);extern struct its_baser *its_lookup_baser (int type); +extern void its_enable_defaults (void );#else /* __arm__ */
lib/arm/gic-v3-its.c +89 -0
92 lines not on this page · raw message
Signature · 1 line
2.20.1
Zenghui Yu
reviewerRe: [kvm-unit-tests PATCH v3 08/14] arm/arm64: ITS: its_enable_defaults
Hi Eric,
Quoted 60 lines · Eric Auger
its_enable_defaults()is the top init function that allocates the command queue and all the requested tables (device, collection, lpi config and pending tables), enable LPIs at distributor level and ITS level.
gicv3_enable_defaultsmust be called before.Signed-off-by Eric Auger <[email protected]>---
v2 -> v3:
- introduceits_setup_baserin this patch
- squash "arm/arm64: ITS: Init the command queue" in this patch.--- lib/arm/asm/gic-v3-its.h | 8 ++++ lib/arm/gic-v3-its.c | 89 ++++++++++++++++++++++++++++++++++++++++ 2 files changed, 97 insertions(+)lib/arm/asm/gic-v3-its.h +8
-0@@ -36,6 +36,8 @@ struct its_data {void *base;struct its_typer typer;struct its_baser baser[GITS_BASER_NR_REGS ]; +struct its_cmd_block *cmd_base; +struct its_cmd_block *cmd_write; };extern struct its_data its_data;@@ -88,10 +90,16 @@ extern struct its_data its_data;#define GITS_BASER_TYPE_DEVICE 1#define GITS_BASER_TYPE_COLLECTION 4 + +struct its_cmd_block { +u64 raw_cmd[4 ]; +}; +extern void its_parse_typer (void );extern void its_init (void );extern int its_parse_baser (int i,struct its_baser *baser);extern struct its_baser *its_lookup_baser (int type); +extern void its_enable_defaults (void );#else /* __arm__ */ lib/arm/gic-v3-its.c +3
-0@@ -86,3 +86,92 @@ void its_init(void)its_parse_baser (i, &its_data.baser[i]); } +static void its_setup_baser (int i,struct its_baser *baser) +{ +unsigned long n = (baser->nr_pages * baser->psz) >>PAGE_SHIFT ;
lib/arm/gic-v3-its.c void its_init(void) · line 101
Quoted 10 lines
lib/arm/gic-v3-its.c +10
-0+unsigned long order =is_power_of_2 (n) ?fls (n) :fls (n) +1 ; +u64 val; + + baser->table_addr = (u64 )virt_to_phys (alloc_pages (order)); + + val = ((u64)baser->table_addr | + ((u64)baser->type << GITS_BASER_TYPE_SHIFT) | + ((u64)(baser->esz - 1) << GITS_BASER_ENTRY_SIZE_SHIFT) | + ((baser->nr_pages - 1) << GITS_BASER_PAGES_SHIFT) | + (u64)baser->indirect << 62 |
I haven't seen the 'nr_pages' and 'indirect' are programmed anywhere except in its_parse_baser(). It looks like they're treated as RO (but they shouldn't) and I now don't think it makes sense to parse them in its_parse_baser(), in patch#5.
Quoted 10 lines
lib/arm/gic-v3-its.c +10
-0+ (u64)baser->valid << 63); + +switch (baser->psz) { +case SZ_4K : + val |=GITS_BASER_PAGE_SIZE_4K ; +break ; +case SZ_16K : + val |=GITS_BASER_PAGE_SIZE_16K ; +break ; +case SZ_64K :
lib/arm/gic-v3-its.c void its_init(void) · line 121
Quoted 10 lines
lib/arm/gic-v3-its.c +10
-0+ val |=GITS_BASER_PAGE_SIZE_64K ; +break ; + } + +writeq (val,gicv3_its_base () +GITS_BASER + i *8 ); +} + +/** +* init_cmd_queue: Allocate the command queue and initialize +* CBASER, CREADR, CWRITER
no 'CREADR'.
Thanks,
Zenghui
Re: [kvm-unit-tests PATCH v3 08/14] arm/arm64: ITS: its_enable_defaults
Hi Zenghui,
Quoted 65 lines · Zenghui Yu
Hi Eric,
Quoted 62 lines · Eric Auger
its_enable_defaults()is the top init function that allocates the command queue and all the requested tables (device, collection, lpi config and pending tables), enable LPIs at distributor level and ITS level.
gicv3_enable_defaultsmust be called before.Signed-off-by Eric Auger <[email protected]>---
v2 -> v3:
- introduceits_setup_baserin this patch
- squash "arm/arm64: ITS: Init the command queue" in this patch.
---
lib/arm/asm/gic-v3-its.h| 8 ++++
lib/arm/gic-v3-its.c | 89 ++++++++++++++++++++++++++++++++++++++++2 files changed, 97 insertions(+)lib/arm/asm/gic-v3-its.h +2
-0@@ -36,6 +36,8 @@ struct its_data { void *base; struct its_typer typer; struct its_baser baser[GITS_BASER_NR_REGS]; + struct its_cmd_block *cmd_base; + struct its_cmd_block *cmd_write; }; extern struct its_data its_data;@@ -88,10 +90,16 @@ extern struct its_data its_data; #define GITS_BASER_TYPE_DEVICE 1 #define GITS_BASER_TYPE_COLLECTION 4+
lib/arm/asm/gic-v3-its.h +4
-0+struct its_cmd_block { + u64 raw_cmd[4]; +}; +extern void
its_parse_typer(void);
extern voidits_init(void);
extern intits_parse_baser(int i, struct its_baser *baser); extern structits_baser*its_lookup_baser(int type); +extern voidits_enable_defaults(void);
#else /*__arm__*/
diff --gita/lib/arm/gic-v3-its.cb/lib/arm/gic-v3-its.cindex 2c0ce13..d1e7e52 100644lib/arm/gic-v3-its.c +9
-0@@ -86,3 +86,92 @@ void its_init(void) its_parse_baser(i, &its_data.baser[i]); } +static void its_setup_baser(int i, struct its_baser *baser) +{ + unsigned long n = (baser->nr_pages * baser->psz) >> PAGE_SHIFT; + unsigned long order = is_power_of_2(n) ? fls(n) : fls(n) + 1; + u64 val; + + baser->table_addr = (u64)virt_to_phys(alloc_pages(order)); + + val = ((u64)baser->table_addr | + ((u64)baser->type << GITS_BASER_TYPE_SHIFT) |
Quoted 8 lines
lib/arm/gic-v3-its.c void its_init(void) · line 101
Quoted 3 lines
lib/arm/gic-v3-its.c +3
-0+ ((u64)(baser->esz - 1) << GITS_BASER_ENTRY_SIZE_SHIFT) | + ((baser->nr_pages - 1) << GITS_BASER_PAGES_SHIFT) | + (u64)baser->indirect << 62 |I haven't seen the '
nr_pages' and 'indirect' are programmed anywhere except inits_parse_baser(). It looks like they're treated as RO (but they shouldn't) and I now don't think it makes sense to parse them inits_parse_baser(), in patch#5.
First of all please forgive me for the delay.
I agree with you on nr_pages. However indirect also indicates the BASER capability to support or not 2 level tables. So I think it makes sense to read it on init.
Quoted 15 lines
Quoted 15 lines
lib/arm/gic-v3-its.c +14
-0+ (u64)baser->valid << 63); + + switch (baser->psz) { + case SZ_4K: + val |= GITS_BASER_PAGE_SIZE_4K; + break; + case SZ_16K: + val |= GITS_BASER_PAGE_SIZE_16K; + break; + case SZ_64K: + val |= GITS_BASER_PAGE_SIZE_64K; + break; + } +
Quoted 8 lines
lib/arm/gic-v3-its.c void its_init(void) · line 121
Quoted 6 lines
lib/arm/gic-v3-its.c +6
-0+ writeq(val, gicv3_its_base() + GITS_BASER + i * 8); +} + +/** +* init_cmd_queue: Allocate the command queue and initialize +* CBASER, CREADR, CWRITER no 'CREADR'.
OK
Thanks
Eric
Quoted 5 lines
Thanks,
Zenghui
Zenghui Yu
reviewerRe: [kvm-unit-tests PATCH v3 08/14] arm/arm64: ITS: its_enable_defaults
Hi Eric,
Quoted 76 lines · Auger Eric
Hi Zenghui,
Quoted 65 lines · Zenghui Yu
Hi Eric,
Quoted 62 lines · Eric Auger
its_enable_defaults()is the top init function that allocates the command queue and all the requested tables (device, collection, lpi config and pending tables), enable LPIs at distributor level and ITS level.
gicv3_enable_defaultsmust be called before.Signed-off-by Eric Auger <[email protected]>---
v2 -> v3:
- introduceits_setup_baserin this patch
- squash "arm/arm64: ITS: Init the command queue" in this patch.--- lib/arm/asm/gic-v3-its.h | 8 ++++ lib/arm/gic-v3-its.c | 89 ++++++++++++++++++++++++++++++++++++++++ 2 files changed, 97 insertions(+)lib/arm/asm/gic-v3-its.h +7
-0@@ -36,6 +36,8 @@ struct its_data { void *base; struct its_typer typer; struct its_baser baser[GITS_BASER_NR_REGS]; + struct its_cmd_block *cmd_base; + struct its_cmd_block *cmd_write; }; extern struct its_data its_data;@@ -88,10 +90,16 @@ extern struct its_data its_data; #define GITS_BASER_TYPE_DEVICE 1 #define GITS_BASER_TYPE_COLLECTION 4 + +struct its_cmd_block { + u64 raw_cmd[4]; +}; + extern void its_parse_typer(void); extern void its_init(void); extern int its_parse_baser(int i, struct its_baser *baser); extern struct its_baser *its_lookup_baser(int type); +extern void its_enable_defaults (void ); #else /* __arm__ */ diff --git a/lib/arm/gic-v3-its.c b/lib/arm/gic-v3-its.cindex 2c0ce13..d1e7e52 100644
lib/arm/gic-v3-its.c +9
-0@@ -86,3 +86,92 @@ void its_init(void) its_parse_baser(i, &its_data.baser[i]); } +static void its_setup_baser(int i, struct its_baser *baser) +{ + unsigned long n = (baser->nr_pages * baser->psz) >> PAGE_SHIFT; + unsigned long order = is_power_of_2(n) ? fls(n) : fls(n) + 1; + u64 val; + + baser->table_addr = (u64)virt_to_phys(alloc_pages(order)); + + val = ((u64)baser->table_addr | + ((u64)baser->type << GITS_BASER_TYPE_SHIFT) |Quoted 8 lines
lib/arm/gic-v3-its.c void its_init(void) · line 101
Quoted 3 lines
lib/arm/gic-v3-its.c +3
-0+ ((u64)(baser->esz - 1) << GITS_BASER_ENTRY_SIZE_SHIFT) | + ((baser->nr_pages - 1) << GITS_BASER_PAGES_SHIFT) | + (u64)baser->indirect << 62 |I haven't seen the '
nr_pages' and 'indirect' are programmed anywhere except inits_parse_baser(). It looks like they're treated as RO (but they shouldn't) and I now don't think it makes sense to parse them inits_parse_baser(), in patch#5.
Quoted 1 line
First of all please forgive me for the delay.
Never mind.
Quoted 4 lines
I agree with you on
nr_pages. However indirect also indicates the BASER capability to support or not 2 level tables. So I think it makes sense to read it on init.
Yes, you're right. As the spec says, the Indirect field "is RAZ/WI for GIC implementations that only support flat tables".
Thanks,
Zenghui
Re: [kvm-unit-tests PATCH v3 08/14] arm/arm64: ITS: its_enable_defaults
In reply to Eric AugerQuoted 61 lines · Eric Auger
its_enable_defaults()is the top init function that allocates the command queue and all the requested tables (device, collection, lpi config and pending tables), enable LPIs at distributor level and ITS level.
gicv3_enable_defaultsmust be called before.Signed-off-by Eric Auger <[email protected]>---
v2 -> v3:
- introduceits_setup_baserin this patch
- squash "arm/arm64: ITS: Init the command queue" in this patch.--- lib/arm/asm/gic-v3-its.h | 8 ++++ lib/arm/gic-v3-its.c | 89 ++++++++++++++++++++++++++++++++++++++++ 2 files changed, 97 insertions(+)lib/arm/asm/gic-v3-its.h +8
-0@@ -36,6 +36,8 @@ struct its_data {void *base;struct its_typer typer;struct its_baser baser[GITS_BASER_NR_REGS ]; +struct its_cmd_block *cmd_base; +struct its_cmd_block *cmd_write; };extern struct its_data its_data;@@ -88,10 +90,16 @@ extern struct its_data its_data;#define GITS_BASER_TYPE_DEVICE 1#define GITS_BASER_TYPE_COLLECTION 4 + +struct its_cmd_block { +u64 raw_cmd[4 ]; +}; +extern void its_parse_typer (void );extern void its_init (void );extern int its_parse_baser (int i,struct its_baser *baser);extern struct its_baser *its_lookup_baser (int type); +extern void its_enable_defaults (void );#else /* __arm__ */ lib/arm/gic-v3-its.c +4
-0@@ -86,3 +86,92 @@ void its_init(void)its_parse_baser (i, &its_data.baser[i]); } +static void its_setup_baser (int i,struct its_baser *baser) +{ +unsigned long n = (baser->nr_pages * baser->psz) >>PAGE_SHIFT ; +unsigned long order =is_power_of_2 (n) ?fls (n) :fls (n) +1 ;
lib/arm/gic-v3-its.c void its_init(void) · line 102
Quoted 10 lines
lib/arm/gic-v3-its.c +10
-0+u64 val; + + baser->table_addr = (u64 )virt_to_phys (alloc_pages (order)); + + val = ((u64 )baser->table_addr | + ((u64 )baser->type <<GITS_BASER_TYPE_SHIFT ) | + ((u64)(baser->esz -1 ) <<GITS_BASER_ENTRY_SIZE_SHIFT ) | + ((baser->nr_pages -1 ) <<GITS_BASER_PAGES_SHIFT ) | + (u64 )baser->indirect <<62 | + (u64 )baser->valid <<63 );
<< GITS_BASER_INDIRECT
<< GITS_BASER_VALID
Quoted 40 lines
lib/arm/gic-v3-its.c +40
-0+ +switch (baser->psz) { +case SZ_4K : + val |=GITS_BASER_PAGE_SIZE_4K ; +break ; +case SZ_16K : + val |=GITS_BASER_PAGE_SIZE_16K ; +break ; +case SZ_64K : + val |=GITS_BASER_PAGE_SIZE_64K ; +break ; + } + +writeq (val,gicv3_its_base () +GITS_BASER + i *8 ); +} + +/** +* init_cmd_queue: Allocate the command queue and initialize +* CBASER, CREADR, CWRITER +*/ +static void its_cmd_queue_init (void ) +{ +unsigned long n =SZ_64K >>PAGE_SHIFT ; +unsigned long order =fls (n); +u64 cbaser; + + its_data.cmd_base = (void *)virt_to_phys (alloc_pages (order)); + + cbaser = ((u64 )its_data.cmd_base | (SZ_64K /SZ_4K -1 ) |GITS_CBASER_VALID ); + +writeq (cbaser, its_data.base +GITS_CBASER ); + + its_data.cmd_write = its_data.cmd_base; +writeq (0 , its_data.base +GITS_CWRITER ); +} + +void its_enable_defaults (void ) +{ +unsigned int i; +
lib/arm/gic-v3-its.c void its_init(void) · line 152
Quoted 10 lines
lib/arm/gic-v3-its.c +10
-0+its_parse_typer (); + +/* Allocate BASER tables (device and collection tables) */ +for (i =0 ; i <GITS_BASER_NR_REGS ; i++) { +struct its_baser *baser = &its_data.baser[i]; +int ret; + + ret =its_parse_baser (i, baser); +if (ret) +continue ;
Didn't we already parse typer and baser at its_init time? How/why would its_parse_baser fail? Should we assert when it does?
lib/arm/gic-v3-its.c void its_init(void) · line 164
Quoted 12 lines
lib/arm/gic-v3-its.c +12
-0+ +switch (baser->type) { +case GITS_BASER_TYPE_DEVICE : + baser->valid = true; +its_setup_baser (i, baser); +break ; +case GITS_BASER_TYPE_COLLECTION : + baser->valid = true; +its_setup_baser (i, baser); +break ; +default : +break ;
assert() ?
Quoted 17 lines
lib/arm/gic-v3-its.c +13
-0+ } + } + +/* Allocate LPI config and pending tables */ +gicv3_lpi_alloc_tables (); + +its_cmd_queue_init (); + +for (i =0 ; i < nr_cpus; i++) +gicv3_lpi_rdist_ctrl (i, true); + +writel (GITS_CTLR_ENABLE , its_data.base +GITS_CTLR ); +}Signature · 1 line
2.20.1
It's looking like we don't have a clean separation between its_init and its_enable_defaults. I'd expect its_init to do all the allocating of memory and its_enable_defaults to do all the write's to the device. We should only do its_init once and its_enable should be something we can do again (after a disable, on reset, etc.). Is that not possible with the ITS device?
Thanks,
drew
Re: [kvm-unit-tests PATCH v3 08/14] arm/arm64: ITS: its_enable_defaults
Hi Drew,
Quoted 67 lines · Andrew Jones
Quoted 66 lines · Eric Auger
its_enable_defaults()is the top init function that allocates the command queue and all the requested tables (device, collection, lpi config and pending tables), enable LPIs at distributor level and ITS level.
gicv3_enable_defaultsmust be called before.Signed-off-by Eric Auger <[email protected]>---
v2 -> v3:
- introduceits_setup_baserin this patch
- squash "arm/arm64: ITS: Init the command queue" in this patch.--- lib/arm/asm/gic-v3-its.h | 8 ++++ lib/arm/gic-v3-its.c | 89 ++++++++++++++++++++++++++++++++++++++++ 2 files changed, 97 insertions(+)lib/arm/asm/gic-v3-its.h +8
-0@@ -36,6 +36,8 @@ struct its_data {void *base;struct its_typer typer;struct its_baser baser[GITS_BASER_NR_REGS ]; +struct its_cmd_block *cmd_base; +struct its_cmd_block *cmd_write; };extern struct its_data its_data;@@ -88,10 +90,16 @@ extern struct its_data its_data;#define GITS_BASER_TYPE_DEVICE 1#define GITS_BASER_TYPE_COLLECTION 4 + +struct its_cmd_block { +u64 raw_cmd[4 ]; +}; +extern void its_parse_typer (void );extern void its_init (void );extern int its_parse_baser (int i,struct its_baser *baser);extern struct its_baser *its_lookup_baser (int type); +extern void its_enable_defaults (void );#else /* __arm__ */ lib/arm/gic-v3-its.c +9
-0@@ -86,3 +86,92 @@ void its_init(void)its_parse_baser (i, &its_data.baser[i]); } +static void its_setup_baser (int i,struct its_baser *baser) +{ +unsigned long n = (baser->nr_pages * baser->psz) >>PAGE_SHIFT ; +unsigned long order =is_power_of_2 (n) ?fls (n) :fls (n) +1 ; +u64 val; + + baser->table_addr = (u64 )virt_to_phys (alloc_pages (order)); + + val = ((u64)baser->table_addr |
Quoted 8 lines
lib/arm/gic-v3-its.c void its_init(void) · line 102
Quoted 5 lines
lib/arm/gic-v3-its.c +5
-0+ ((u64 )baser->type <<GITS_BASER_TYPE_SHIFT ) | + ((u64)(baser->esz -1 ) <<GITS_BASER_ENTRY_SIZE_SHIFT ) | + ((baser->nr_pages -1 ) <<GITS_BASER_PAGES_SHIFT ) | + (u64 )baser->indirect <<62 | + (u64)baser->valid << 63);<<
GITS_BASER_INDIRECT
<<GITS_BASER_VALID
Those are BIT()
if (baser->indirect)
val |= GITS_BASER_INDIRECT;
if (baser->valid)
val |= GITS_BASER_VALID;Quoted 46 lines
Quoted 46 lines
lib/arm/gic-v3-its.c +45
-0+ +switch (baser->psz) { +case SZ_4K : + val |=GITS_BASER_PAGE_SIZE_4K ; +break ; +case SZ_16K : + val |=GITS_BASER_PAGE_SIZE_16K ; +break ; +case SZ_64K : + val |=GITS_BASER_PAGE_SIZE_64K ; +break ; + } + +writeq (val,gicv3_its_base () +GITS_BASER + i *8 ); +} + +/** +* init_cmd_queue: Allocate the command queue and initialize +* CBASER, CREADR, CWRITER +*/ +static void its_cmd_queue_init (void ) +{ +unsigned long n =SZ_64K >>PAGE_SHIFT ; +unsigned long order =fls (n); +u64 cbaser; + + its_data.cmd_base = (void *)virt_to_phys (alloc_pages (order)); + + cbaser = ((u64 )its_data.cmd_base | (SZ_64K /SZ_4K -1 ) |GITS_CBASER_VALID ); + +writeq (cbaser, its_data.base +GITS_CBASER ); + + its_data.cmd_write = its_data.cmd_base; +writeq (0 , its_data.base +GITS_CWRITER ); +} + +void its_enable_defaults (void ) +{ +unsigned int i; + +its_parse_typer (); + +/* Allocate BASER tables (device and collection tables) */ +for (i =0 ; i <GITS_BASER_NR_REGS ; i++) { +struct its_baser *baser = &its_data.baser[i];
Quoted 8 lines
lib/arm/gic-v3-its.c void its_init(void) · line 152
Quoted 5 lines
lib/arm/gic-v3-its.c +5
-0+int ret; + + ret =its_parse_baser (i, baser); +if (ret) +continue ;Didn't we already parse typer and baser at
its_inittime? How/why wouldits_parse_baserfail? Should we assert when it does?
Yes we parsed them already so I can directly use
baser = &its_data.baser[i];
More generally I have simplified and only cares about the device and
collection baser now.
Quoted 32 lines
lib/arm/gic-v3-its.c void its_init(void) · line 164
Quoted 13 lines
lib/arm/gic-v3-its.c +12
-0+ +switch (baser->type) { +case GITS_BASER_TYPE_DEVICE : + baser->valid = true; +its_setup_baser (i, baser); +break ; +case GITS_BASER_TYPE_COLLECTION : + baser->valid = true; +its_setup_baser (i, baser); +break ; +default : +break ;
assert()?Quoted 16 lines
lib/arm/gic-v3-its.c +13
-0+ } + } + +/* Allocate LPI config and pending tables */ +gicv3_lpi_alloc_tables (); + +its_cmd_queue_init (); + +for (i =0 ; i < nr_cpus; i++) +gicv3_lpi_rdist_ctrl (i, true); + +writel (GITS_CTLR_ENABLE , its_data.base +GITS_CTLR ); +}Signature · 1 line
2.20.1
Quoted 8 lines
Quoted 1 line
It's looking like we don't have a clean separation between
its_initandits_enable_defaults. I'd expectits_initto do all the allocating of memory andits_enable_defaultsto do all the write's to the device. We should only doits_initonce andits_enableshould be something we can do again (after a disable, on reset, etc.). Is that not possible with the ITS device?
I can moved the BASER device & collection table allocation, + the queue init in its_init()
However gicv3_lpi_alloc_tables() must be called after gicv3_enable_defaults as it uses the redist_base[] initialized there.
Thanks
Eric
Quoted 4 lines
Thanks,
drew
_______________________________________________
kvmarm mailing list
[email protected]
https://lists.cs.columbia.edu/mailman/listinfo/kvmarm
PATCH v3 9/14 arm/arm64: ITS: Device and collection Initialization
In reply to Eric AugerIntroduce an helper functions to register
- a new device, characterized by its device id and the max number of event IDs that dimension its ITT (Interrupt Translation Table). The function allocates the ITT.
- a new collection, characterized by its ID and the target processing engine (PE).
---
v2 -> v3:
- s/report_abort/assert
v1 -> v2:
- s/nb_/nr_
--- lib/arm/asm/gic-v3-its.h | 20 +++++++++++++++++- lib/arm/gic-v3-its.c | 44 ++++++++++++++++++++++++++++++++++++++++ 2 files changed, 63 insertions(+), 1 deletion(-)
lib/arm/asm/gic-v3-its.h +19 -1
@@ -31,6 +31,19 @@ struct its_baser { };#define GITS_BASER_NR_REGS 8 +#define GITS_MAX_DEVICES 8 +#define GITS_MAX_COLLECTIONS 8 + +struct its_device { +u32 device_id;/* device ID */ +u32 nr_ites;/* Max Interrupt Translation Entries */ +void *itt;/* Interrupt Translation Table GPA */ +}; + +struct its_collection { +u64 target_address; +u16 col_id; +};struct its_data {void *base;
@@ -38,6 +51,10 @@ struct its_data {struct its_baser baser[GITS_BASER_NR_REGS ];struct its_cmd_block *cmd_base;struct its_cmd_block *cmd_write; +struct its_device devices[GITS_MAX_DEVICES ]; +u32 nr_devices;/* Allocated Devices */ +struct its_collection collections[GITS_MAX_COLLECTIONS ]; +u32 nr_collections;/* Allocated Collections */ };extern struct its_data its_data;
@@ -90,7 +107,6 @@ extern struct its_data its_data;#define GITS_BASER_TYPE_DEVICE 1#define GITS_BASER_TYPE_COLLECTION 4 -struct its_cmd_block {u64 raw_cmd[4 ]; };
@@ -100,6 +116,8 @@ extern void its_init(void);extern int its_parse_baser (int i,struct its_baser *baser);extern struct its_baser *its_lookup_baser (int type);extern void its_enable_defaults (void ); +extern struct its_device *its_create_device (u32 dev_id,int nr_ites); +extern struct its_collection *its_create_collection (u32 col_id,u32 target_pe);#else /* __arm__ */
lib/arm/gic-v3-its.c +44 -0
47 lines not on this page · raw message
Signature · 1 line
2.20.1
Zenghui Yu
reviewerRe: [kvm-unit-tests PATCH v3 09/14] arm/arm64: ITS: Device and collection Initialization
Hi Eric,
Quoted 86 lines · Eric Auger
Introduce an helper functions to register
- a new device, characterized by its device id and the max number of event IDs that dimension its ITT (Interrupt Translation Table). The function allocates the ITT.- a new collection, characterized by its ID and the target processing engine (PE).
Signed-off-by Eric Auger <[email protected]>---
v2 -> v3:
- s/report_abort/assertv1 -> v2:
- s/nb_/nr_--- lib/arm/asm/gic-v3-its.h | 20 +++++++++++++++++- lib/arm/gic-v3-its.c | 44 ++++++++++++++++++++++++++++++++++++++++ 2 files changed, 63 insertions(+), 1 deletion(-)lib/arm/asm/gic-v3-its.h +19
-1@@ -31,6 +31,19 @@ struct its_baser { };#define GITS_BASER_NR_REGS 8 +#define GITS_MAX_DEVICES 8 +#define GITS_MAX_COLLECTIONS 8 + +struct its_device { +u32 device_id;/* device ID */ +u32 nr_ites;/* Max Interrupt Translation Entries */ +void *itt;/* Interrupt Translation Table GPA */ +}; + +struct its_collection { +u64 target_address; +u16 col_id; +};struct its_data {void *base;@@ -38,6 +51,10 @@ struct its_data {struct its_baser baser[GITS_BASER_NR_REGS ];struct its_cmd_block *cmd_base;struct its_cmd_block *cmd_write; +struct its_device devices[GITS_MAX_DEVICES ]; +u32 nr_devices;/* Allocated Devices */ +struct its_collection collections[GITS_MAX_COLLECTIONS ]; +u32 nr_collections;/* Allocated Collections */ };extern struct its_data its_data;@@ -90,7 +107,6 @@ extern struct its_data its_data;#define GITS_BASER_TYPE_DEVICE 1#define GITS_BASER_TYPE_COLLECTION 4 -struct its_cmd_block {u64 raw_cmd[4 ]; };@@ -100,6 +116,8 @@ extern void its_init(void);extern int its_parse_baser (int i,struct its_baser *baser);extern struct its_baser *its_lookup_baser (int type);extern void its_enable_defaults (void ); +extern struct its_device *its_create_device (u32 dev_id,int nr_ites); +extern struct its_collection *its_create_collection (u32 col_id,u32 target_pe);#else /* __arm__ */ lib/arm/gic-v3-its.c +4
-0@@ -175,3 +175,47 @@ void its_enable_defaults(void)writel (GITS_CTLR_ENABLE , its_data.base +GITS_CTLR ); } + +struct its_device *its_create_device (u32 device_id,int nr_ites) +{ +struct its_baser *baser;
lib/arm/gic-v3-its.c void its_enable_defaults(void) · line 189
Quoted 10 lines
lib/arm/gic-v3-its.c +8
-0+struct its_device *new; +unsigned long n, order; + +assert (its_data.nr_devices <GITS_MAX_DEVICES ); + + baser =its_lookup_baser (GITS_BASER_TYPE_DEVICE ); +if (!baser) +return NULL ;
I think there's no need to lookup the device baser here. As the device baser should have already been setup at initialization time (i.e. in its_enable_defaults). And anyway, 'baser' is not being used in this function.
Thanks,
Zenghui
Quoted 33 lines
lib/arm/gic-v3-its.c +32
-0+ + new = &its_data.devices[its_data.nr_devices]; + + new->device_id = device_id; + new->nr_ites = nr_ites; + + n = (its_data.typer.ite_size * nr_ites) >>PAGE_SHIFT ; + order =is_power_of_2 (n) ?fls (n) :fls (n) +1 ; + new->itt = (void *)virt_to_phys (alloc_pages (order)); + + its_data.nr_devices++; +return new; +} + +struct its_collection *its_create_collection (u32 col_id,u32 pe) +{ +struct its_collection *new; + +assert (its_data.nr_collections <GITS_MAX_COLLECTIONS ); + + new = &its_data.collections[its_data.nr_collections]; + + new->col_id = col_id; + +if (its_data.typer.pta) + new->target_address = (u64 )gicv3_data.redist_base[pe]; +else + new->target_address = pe <<16 ; + + its_data.nr_collections++; +return new; +}
Re: [kvm-unit-tests PATCH v3 09/14] arm/arm64: ITS: Device and collection Initialization
Hi Zenghui,
Quoted 90 lines · Zenghui Yu
Hi Eric,
Quoted 85 lines · Eric Auger
Introduce an helper functions to register
- a new device, characterized by its device id and the max number of event IDs that dimension its ITT (Interrupt Translation Table). The function allocates the ITT.- a new collection, characterized by its ID and the target processing engine (PE).
Signed-off-by Eric Auger <[email protected]>---
v2 -> v3:
- s/report_abort/assertv1 -> v2:
- s/nb_/nr_
---
lib/arm/asm/gic-v3-its.h| 20 +++++++++++++++++- lib/arm/gic-v3-its.c | 44 ++++++++++++++++++++++++++++++++++++++++2 files changed, 63 insertions(+), 1 deletion(-)lib/arm/asm/gic-v3-its.h +17
-0@@ -31,6 +31,19 @@ struct its_baser { }; #define GITS_BASER_NR_REGS 8 +#define GITS_MAX_DEVICES 8 +#define GITS_MAX_COLLECTIONS 8 + +struct its_device { + u32 device_id; /* device ID */ + u32 nr_ites; /* Max Interrupt Translation Entries */ + void *itt; /* Interrupt Translation Table GPA */ +}; + +struct its_collection { + u64 target_address; + u16 col_id; +}; struct its_data { void *base;@@ -38,6 +51,10 @@ struct its_data { struct its_baser baser[GITS_BASER_NR_REGS]; struct its_cmd_block *cmd_base; struct its_cmd_block *cmd_write; + struct its_device devices[GITS_MAX_DEVICES]; + u32 nr_devices; /* Allocated Devices */ + struct its_collection collections[GITS_MAX_COLLECTIONS]; + u32 nr_collections; /* Allocated Collections */ }; extern struct its_data its_data;@@ -90,7 +107,6 @@ extern struct its_data its_data; #define GITS_BASER_TYPE_DEVICE 1 #define GITS_BASER_TYPE_COLLECTION 4-
structits_cmd_block{
u64 raw_cmd[4];
};lib/arm/asm/gic-v3-its.h +2
-0@@ -100,6 +116,8 @@ extern void its_init(void); extern int its_parse_baser(int i, struct its_baser *baser); extern struct its_baser *its_lookup_baser(int type); extern void its_enable_defaults(void); +extern struct its_device *its_create_device (u32 dev_id,int nr_ites); +extern struct its_collection *its_create_collection (u32 col_id,u32 target_pe); #else /* __arm__ */diff --git
a/lib/arm/gic-v3-its.cb/lib/arm/gic-v3-its.cindex d1e7e52..c2dcd01 100644lib/arm/gic-v3-its.c +9
-0@@ -175,3 +175,47 @@ void its_enable_defaults(void) writel(GITS_CTLR_ENABLE, its_data.base + GITS_CTLR); } + +struct its_device *its_create_device (u32 device_id,int nr_ites) +{ + struct its_baser *baser; + struct its_device *new; + unsigned long n, order; + + assert(its_data.nr_devices < GITS_MAX_DEVICES); +
Quoted 8 lines
lib/arm/gic-v3-its.c void its_enable_defaults(void) · line 189
Quoted 3 lines
lib/arm/gic-v3-its.c +3
-0+ baser = its_lookup_baser(GITS_BASER_TYPE_DEVICE); + if (!baser) + return NULL;I think there's no need to lookup the device baser here. As the device baser should have already been setup at initialization time (i.e. in
its_enable_defaults). And anyway, 'baser' is not being used in this function.
I fully agree. I Removed that.
Thanks!
Eric
Quoted 39 lines
Thanks,
ZenghuiQuoted 33 lines
lib/arm/gic-v3-its.c +32
-0+ + new = &its_data.devices[its_data.nr_devices]; + + new->device_id = device_id; + new->nr_ites = nr_ites; + + n = (its_data.typer.ite_size * nr_ites) >> PAGE_SHIFT; + order = is_power_of_2(n) ? fls(n) : fls(n) + 1; + new->itt = (void *)virt_to_phys(alloc_pages(order)); + + its_data.nr_devices++; + return new; +} + +struct its_collection *its_create_collection (u32 col_id,u32 pe) +{ + struct its_collection *new; + + assert(its_data.nr_collections < GITS_MAX_COLLECTIONS); + + new = &its_data.collections[its_data.nr_collections]; + + new->col_id = col_id; + + if (its_data.typer.pta) + new->target_address = (u64)gicv3_data.redist_base[pe]; + else + new->target_address = pe << 16; + + its_data.nr_collections++; + return new; +}
Re: [kvm-unit-tests PATCH v3 09/14] arm/arm64: ITS: Device and collection Initialization
In reply to Eric AugerQuoted 84 lines · Eric Auger
Introduce an helper functions to register
- a new device, characterized by its device id and the max number of event IDs that dimension its ITT (Interrupt Translation Table). The function allocates the ITT.- a new collection, characterized by its ID and the target processing engine (PE).
Signed-off-by Eric Auger <[email protected]>---
v2 -> v3:
- s/report_abort/assertv1 -> v2:
- s/nb_/nr_--- lib/arm/asm/gic-v3-its.h | 20 +++++++++++++++++- lib/arm/gic-v3-its.c | 44 ++++++++++++++++++++++++++++++++++++++++ 2 files changed, 63 insertions(+), 1 deletion(-)lib/arm/asm/gic-v3-its.h +19
-1@@ -31,6 +31,19 @@ struct its_baser { };#define GITS_BASER_NR_REGS 8 +#define GITS_MAX_DEVICES 8 +#define GITS_MAX_COLLECTIONS 8 + +struct its_device { +u32 device_id;/* device ID */ +u32 nr_ites;/* Max Interrupt Translation Entries */ +void *itt;/* Interrupt Translation Table GPA */ +}; + +struct its_collection { +u64 target_address; +u16 col_id; +};struct its_data {void *base;@@ -38,6 +51,10 @@ struct its_data {struct its_baser baser[GITS_BASER_NR_REGS ];struct its_cmd_block *cmd_base;struct its_cmd_block *cmd_write; +struct its_device devices[GITS_MAX_DEVICES ]; +u32 nr_devices;/* Allocated Devices */ +struct its_collection collections[GITS_MAX_COLLECTIONS ]; +u32 nr_collections;/* Allocated Collections */ };extern struct its_data its_data;@@ -90,7 +107,6 @@ extern struct its_data its_data;#define GITS_BASER_TYPE_DEVICE 1#define GITS_BASER_TYPE_COLLECTION 4 -struct its_cmd_block {u64 raw_cmd[4 ]; };@@ -100,6 +116,8 @@ extern void its_init(void);extern int its_parse_baser (int i,struct its_baser *baser);extern struct its_baser *its_lookup_baser (int type);extern void its_enable_defaults (void ); +extern struct its_device *its_create_device (u32 dev_id,int nr_ites); +extern struct its_collection *its_create_collection (u32 col_id,u32 target_pe);#else /* __arm__ */ lib/arm/gic-v3-its.c +2
-0@@ -175,3 +175,47 @@ void its_enable_defaults(void)writel (GITS_CTLR_ENABLE , its_data.base +GITS_CTLR ); } + +struct its_device *its_create_device (u32 device_id,int nr_ites)
lib/arm/gic-v3-its.c void its_enable_defaults(void) · line 189
Quoted 10 lines
lib/arm/gic-v3-its.c +10
-0+{ +struct its_baser *baser; +struct its_device *new; +unsigned long n, order; + +assert (its_data.nr_devices <GITS_MAX_DEVICES ); + + baser =its_lookup_baser (GITS_BASER_TYPE_DEVICE ); +if (!baser) +return NULL ;
Should we assert here if we can't find a GITS_BASER_TYPE_DEVICE ? It seems none of the callers of its_create_device are checking for null.
lib/arm/gic-v3-its.c void its_enable_defaults(void) · line 197
Quoted 8 lines
lib/arm/gic-v3-its.c +8
-0+ + new = &its_data.devices[its_data.nr_devices]; + + new->device_id = device_id; + new->nr_ites = nr_ites; + + n = (its_data.typer.ite_size * nr_ites) >>PAGE_SHIFT ; + order =is_power_of_2 (n) ?fls (n) :fls (n) +1 ;
I've seen this calculation several times now, so I think an arch-neutral order calculator is in order:
int get_order(size_t size);
lib/arm/gic-v3-its.c void its_enable_defaults(void) · line 198
Quoted 1 line
lib/arm/gic-v3-its.c +1
-0+ new->itt = (void *)virt_to_phys (alloc_pages (order));
If this is a physical address then shouldn't itt be phys_addr_t ?
Quoted 26 lines
lib/arm/gic-v3-its.c +23
-0+ + its_data.nr_devices++; +return new; +} + +struct its_collection *its_create_collection (u32 col_id,u32 pe) +{ +struct its_collection *new; + +assert (its_data.nr_collections <GITS_MAX_COLLECTIONS ); + + new = &its_data.collections[its_data.nr_collections]; + + new->col_id = col_id; + +if (its_data.typer.pta) + new->target_address = (u64 )gicv3_data.redist_base[pe]; +else + new->target_address = pe <<16 ; + + its_data.nr_collections++; +return new; +}Signature · 1 line
2.20.1
Thanks,
drew
Re: [kvm-unit-tests PATCH v3 09/14] arm/arm64: ITS: Device and collection Initialization
Hi Drew,
Quoted 91 lines · Andrew Jones
Quoted 90 lines · Eric Auger
Introduce an helper functions to register
- a new device, characterized by its device id and the max number of event IDs that dimension its ITT (Interrupt Translation Table). The function allocates the ITT.- a new collection, characterized by its ID and the target processing engine (PE).
Signed-off-by Eric Auger <[email protected]>---
v2 -> v3:
- s/report_abort/assertv1 -> v2:
- s/nb_/nr_--- lib/arm/asm/gic-v3-its.h | 20 +++++++++++++++++- lib/arm/gic-v3-its.c | 44 ++++++++++++++++++++++++++++++++++++++++ 2 files changed, 63 insertions(+), 1 deletion(-)lib/arm/asm/gic-v3-its.h +19
-1@@ -31,6 +31,19 @@ struct its_baser { };#define GITS_BASER_NR_REGS 8 +#define GITS_MAX_DEVICES 8 +#define GITS_MAX_COLLECTIONS 8 + +struct its_device { +u32 device_id;/* device ID */ +u32 nr_ites;/* Max Interrupt Translation Entries */ +void *itt;/* Interrupt Translation Table GPA */ +}; + +struct its_collection { +u64 target_address; +u16 col_id; +};struct its_data {void *base;@@ -38,6 +51,10 @@ struct its_data {struct its_baser baser[GITS_BASER_NR_REGS ];struct its_cmd_block *cmd_base;struct its_cmd_block *cmd_write; +struct its_device devices[GITS_MAX_DEVICES ]; +u32 nr_devices;/* Allocated Devices */ +struct its_collection collections[GITS_MAX_COLLECTIONS ]; +u32 nr_collections;/* Allocated Collections */ };extern struct its_data its_data;@@ -90,7 +107,6 @@ extern struct its_data its_data;#define GITS_BASER_TYPE_DEVICE 1#define GITS_BASER_TYPE_COLLECTION 4 -struct its_cmd_block {u64 raw_cmd[4 ]; };@@ -100,6 +116,8 @@ extern void its_init(void);extern int its_parse_baser (int i,struct its_baser *baser);extern struct its_baser *its_lookup_baser (int type);extern void its_enable_defaults (void ); +extern struct its_device *its_create_device (u32 dev_id,int nr_ites); +extern struct its_collection *its_create_collection (u32 col_id,u32 target_pe);#else /* __arm__ */ lib/arm/gic-v3-its.c +8
-0@@ -175,3 +175,47 @@ void its_enable_defaults(void)writel (GITS_CTLR_ENABLE , its_data.base +GITS_CTLR ); } + +struct its_device *its_create_device (u32 device_id,int nr_ites) +{ +struct its_baser *baser; +struct its_device *new; +unsigned long n, order; + +assert (its_data.nr_devices <GITS_MAX_DEVICES );
Quoted 8 lines
lib/arm/gic-v3-its.c void its_enable_defaults(void) · line 189
Quoted 4 lines
lib/arm/gic-v3-its.c +4
-0+ + baser =its_lookup_baser (GITS_BASER_TYPE_DEVICE ); +if (!baser) +return NULL ;Should we assert here if we can't find a
GITS_BASER_TYPE_DEVICE? It seems none of the callers ofits_create_deviceare checking for null.
I removed that code which is useless.
Quoted 13 lines
lib/arm/gic-v3-its.c void its_enable_defaults(void) · line 197
Quoted 9 lines
lib/arm/gic-v3-its.c +8
-0+ + new = &its_data.devices[its_data.nr_devices]; + + new->device_id = device_id; + new->nr_ites = nr_ites; + + n = (its_data.typer.ite_size * nr_ites) >>PAGE_SHIFT ; + order =is_power_of_2 (n) ?fls (n) :fls (n) +1 ;I've seen this calculation several times now, so I think an arch-neutral order calculator is in order:
Quoted 3 lines
int
get_order(size_t size);lib/arm/gic-v3-its.c void its_enable_defaults(void) · line 198
Quoted 1 line
lib/arm/gic-v3-its.c +1
-0+ new->itt = (void *)virt_to_phys (alloc_pages (order));
I added this get_order helper in lib/alloc_page.h/c
Quoted 7 lines
If this is a physical address then shouldn't itt be
phys_addr_t?lib/arm/gic-v3-its.c void its_enable_defaults(void) · line 202
Quoted 4 lines
lib/arm/gic-v3-its.c +4
-0+ + its_data.nr_devices++; +return new; +}
ITT now is a GVA (void *).
Quoted 17 lines
Quoted 17 lines
lib/arm/gic-v3-its.c +17
-0+ +struct its_collection *its_create_collection (u32 col_id,u32 pe) +{ +struct its_collection *new; + +assert (its_data.nr_collections <GITS_MAX_COLLECTIONS ); + + new = &its_data.collections[its_data.nr_collections]; + + new->col_id = col_id; + +if (its_data.typer.pta) + new->target_address = (u64 )gicv3_data.redist_base[pe]; +else + new->target_address = pe <<16 ; + + its_data.nr_collections++;
Quoted 9 lines
Quoted 5 lines
lib/arm/gic-v3-its.c +2
-0+return new; +}Signature · 1 line
2.20.1Thanks,
drew
Thanks
Eric
PATCH v3 10/14 arm/arm64: ITS: commands
In reply to Eric AugerImplement main ITS commands. The code is largely inherited from the ITS driver.
---
v2 -> v3:
- do not use report() anymore
- assert if cmd_write exceeds the queue capacity
v1 -> v2:
- removed its_print_cmd_state
--- arm/Makefile.arm64 | 2 +- lib/arm/asm/gic-v3-its.h | 38 +++- lib/arm/gic-v3-its-cmd.c | 454 +++++++++++++++++++++++++++++++++++++++ 3 files changed, 492 insertions(+), 2 deletions(-) create mode 100644 lib/arm/gic-v3-its-cmd.c
arm/Makefile.arm64 +1 -1
@@ -19,7 +19,7 @@ endef cstart.o = $(TEST_DIR)/cstart64.o cflatobjs += lib/arm64/processor.o cflatobjs += lib/arm64/spinlock.o -cflatobjs += lib/arm/gic-v3-its.o +cflatobjs += lib/arm/gic-v3-its.o lib/arm/gic-v3-its-cmd.o OBJDIRS += lib/arm64
lib/arm/asm/gic-v3-its.h +37 -1
@@ -45,6 +45,8 @@ struct its_collection {u16 col_id; }; +struct its_cmd_block ; +struct its_data {void *base;struct its_typer typer;
@@ -107,6 +109,24 @@ extern struct its_data its_data;#define GITS_BASER_TYPE_DEVICE 1#define GITS_BASER_TYPE_COLLECTION 4 +/* +* ITS commands +*/ +#define GITS_CMD_MAPD 0x08 +#define GITS_CMD_MAPC 0x09 +#define GITS_CMD_MAPTI 0x0a +/* older GIC documentation used MAPVI for this command */ +#define GITS_CMD_MAPVI GITS_CMD_MAPTI +#define GITS_CMD_MAPI 0x0b +#define GITS_CMD_MOVI 0x01 +#define GITS_CMD_DISCARD 0x0f +#define GITS_CMD_INV 0x0c +#define GITS_CMD_MOVALL 0x0e +#define GITS_CMD_INVALL 0x0d +#define GITS_CMD_INT 0x03 +#define GITS_CMD_CLEAR 0x04 +#define GITS_CMD_SYNC 0x05 +struct its_cmd_block {u64 raw_cmd[4 ]; };
@@ -119,11 +139,27 @@ extern void its_enable_defaults(void);extern struct its_device *its_create_device (u32 dev_id,int nr_ites);extern struct its_collection *its_create_collection (u32 col_id,u32 target_pe); +extern void its_send_mapd (struct its_device *dev,int valid); +extern void its_send_mapc (struct its_collection *col,int valid); +extern void its_send_mapti (struct its_device *dev,u32 irq_id, +u32 event_id,struct its_collection *col); +extern void its_send_int (struct its_device *dev,u32 event_id); +extern void its_send_inv (struct its_device *dev,u32 event_id); +extern void its_send_discard (struct its_device *dev,u32 event_id); +extern void its_send_clear (struct its_device *dev,u32 event_id); +extern void its_send_invall (struct its_collection *col); +extern void its_send_movi (struct its_device *dev, +struct its_collection *col,u32 id); +extern void its_send_sync (struct its_collection *col); + +#define ITS_FLAGS_CMDQ_NEEDS_FLUSHING (1ULL << 0) +#define ITS_FLAGS_WORKAROUND_CAVIUM_22375 (1ULL << 1) +#define ITS_FLAGS_WORKAROUND_CAVIUM_23144 (1ULL << 2) +#else /* __arm__ */ static inline void its_init (void ) {}#endif -#endif /* !__ASSEMBLY__ */ #endif /* _ASMARM_GIC_V3_ITS_H_ */
lib/arm/gic-v3-its-cmd.c new file +454 -0
454 lines not on this page · raw message
Signature · 1 line
2.20.1
Re: [kvm-unit-tests PATCH v3 10/14] arm/arm64: ITS: commands
Quoted 38 lines · Eric Auger
Implement main ITS commands. The code is largely inherited from the ITS driver.
Signed-off-by Eric Auger <[email protected]>---
v2 -> v3:
- do not usereport()anymore
- assert ifcmd_writeexceeds the queue capacityv1 -> v2:
- removedits_print_cmd_state--- arm/Makefile.arm64 | 2 +- lib/arm/asm/gic-v3-its.h | 38 +++- lib/arm/gic-v3-its-cmd.c | 454 +++++++++++++++++++++++++++++++++++++++ 3 files changed, 492 insertions(+), 2 deletions(-) create mode 100644 lib/arm/gic-v3-its-cmd.carm/Makefile.arm64 +1
-1@@ -19,7 +19,7 @@ endef cstart.o = $(TEST_DIR)/cstart64.o cflatobjs += lib/arm64/processor.o cflatobjs += lib/arm64/spinlock.o -cflatobjs += lib/arm/gic-v3-its.o +cflatobjs += lib/arm/gic-v3-its.o lib/arm/gic-v3-its-cmd.o OBJDIRS += lib/arm64lib/arm/asm/gic-v3-its.h +0
-0
lib/arm/asm/gic-v3-its.h struct its_collection { · line 49
Quoted 5 lines
lib/arm/asm/gic-v3-its.h +2
-0@@ -45,6 +45,8 @@ struct its_collection {u16 col_id; }; +struct its_cmd_block ; +
This isn't necessary. If it was, then it should have been added in a previous patch.
Quoted 14 lines
lib/arm/asm/gic-v3-its.h +7
-0struct its_data {void *base;struct its_typer typer;@@ -107,6 +109,24 @@ extern struct its_data its_data;#define GITS_BASER_TYPE_DEVICE 1#define GITS_BASER_TYPE_COLLECTION 4 +/* +* ITS commands +*/ +#define GITS_CMD_MAPD 0x08 +#define GITS_CMD_MAPC 0x09 +#define GITS_CMD_MAPTI 0x0a +/* older GIC documentation used MAPVI for this command */
lib/arm/asm/gic-v3-its.h extern struct its_data its_data; · line 128
Quoted 10 lines
lib/arm/asm/gic-v3-its.h +10
-0+#define GITS_CMD_MAPVI GITS_CMD_MAPTI +#define GITS_CMD_MAPI 0x0b +#define GITS_CMD_MOVI 0x01 +#define GITS_CMD_DISCARD 0x0f +#define GITS_CMD_INV 0x0c +#define GITS_CMD_MOVALL 0x0e +#define GITS_CMD_INVALL 0x0d +#define GITS_CMD_INT 0x03 +#define GITS_CMD_CLEAR 0x04 +#define GITS_CMD_SYNC 0x05
Please use tabs.
Quoted 14 lines
lib/arm/asm/gic-v3-its.h +7
-0+struct its_cmd_block {u64 raw_cmd[4 ]; };@@ -119,11 +139,27 @@ extern void its_enable_defaults(void);extern struct its_device *its_create_device (u32 dev_id,int nr_ites);extern struct its_collection *its_create_collection (u32 col_id,u32 target_pe); +extern void its_send_mapd (struct its_device *dev,int valid); +extern void its_send_mapc (struct its_collection *col,int valid); +extern void its_send_mapti (struct its_device *dev,u32 irq_id, +u32 event_id,struct its_collection *col); +extern void its_send_int (struct its_device *dev,u32 event_id); +extern void its_send_inv (struct its_device *dev,u32 event_id);
lib/arm/asm/gic-v3-its.h extern void its_enable_defaults(void); · line 157
Quoted 10 lines
lib/arm/asm/gic-v3-its.h +10
-0+extern void its_send_discard (struct its_device *dev,u32 event_id); +extern void its_send_clear (struct its_device *dev,u32 event_id); +extern void its_send_invall (struct its_collection *col); +extern void its_send_movi (struct its_device *dev, +struct its_collection *col,u32 id); +extern void its_send_sync (struct its_collection *col); + +#define ITS_FLAGS_CMDQ_NEEDS_FLUSHING (1ULL << 0) +#define ITS_FLAGS_WORKAROUND_CAVIUM_22375 (1ULL << 1) +#define ITS_FLAGS_WORKAROUND_CAVIUM_23144 (1ULL << 2)
What are these flags for?
Quoted 17 lines
lib/arm/asm/gic-v3-its.h +1
-1+#else /* __arm__ */ static inline void its_init (void ) {}#endif -#endif /* !__ASSEMBLY__ */ #endif /* _ASMARM_GIC_V3_ITS_H_ */ lib/arm/gic-v3-its-cmd.c new file +2
-0@@ -0,0 +1,454 @@ +/* +* Copyright (C) 2020, Red Hat Inc, Eric Auger <[email protected]>
lib/arm/gic-v3-its-cmd.c line 12
Quoted 10 lines
lib/arm/gic-v3-its-cmd.c new file +10
-0+* +* Most of the code is copy-pasted from: +* drivers/irqchip/irq-gic-v3-its.c +* This work is licensed under the terms of the GNU LGPL, version 2. +*/ +#include <asm/io.h> +#include <asm/gic.h> +#include <asm/gic-v3-its.h> + +#define ITS_ITT_ALIGN SZ_256
tabs
Quoted 143 lines
lib/arm/gic-v3-its-cmd.c new file +143
-0+ +static const char *const its_cmd_string[] = { + [GITS_CMD_MAPD ] ="MAPD" , + [GITS_CMD_MAPC ] ="MAPC" , + [GITS_CMD_MAPTI ] ="MAPTI" , + [GITS_CMD_MAPI ] ="MAPI" , + [GITS_CMD_MOVI ] ="MOVI" , + [GITS_CMD_DISCARD ] ="DISCARD" , + [GITS_CMD_INV ] ="INV" , + [GITS_CMD_MOVALL ] ="MOVALL" , + [GITS_CMD_INVALL ] ="INVALL" , + [GITS_CMD_INT ] ="INT" , + [GITS_CMD_CLEAR ] ="CLEAR" , + [GITS_CMD_SYNC ] ="SYNC" , +}; + +struct its_cmd_desc { +union { +struct { +struct its_device *dev; +u32 event_id; + } its_inv_cmd; + +struct { +struct its_device *dev; +u32 event_id; + } its_int_cmd; + +struct { +struct its_device *dev; +bool valid; + } its_mapd_cmd; + +struct { +struct its_collection *col; +bool valid; + } its_mapc_cmd; + +struct { +struct its_device *dev; +u32 phys_id; +u32 event_id; +u32 col_id; + } its_mapti_cmd; + +struct { +struct its_device *dev; +struct its_collection *col; +u32 event_id; + } its_movi_cmd; + +struct { +struct its_device *dev; +u32 event_id; + } its_discard_cmd; + +struct { +struct its_device *dev; +u32 event_id; + } its_clear_cmd; + +struct { +struct its_collection *col; + } its_invall_cmd; + +struct { +struct its_collection *col; + } its_sync_cmd; + }; +}; + +typedef void (*its_cmd_builder_t )(struct its_cmd_block *, +struct its_cmd_desc *); + +/* ITS COMMANDS */ + +static void its_encode_cmd (struct its_cmd_block *cmd,u8 cmd_nr) +{ + cmd->raw_cmd[0 ] &= ~0xffUL ; + cmd->raw_cmd[0 ] |= cmd_nr; +} + +static void its_encode_devid (struct its_cmd_block *cmd,u32 devid) +{ + cmd->raw_cmd[0 ] &=BIT_ULL (32 ) -1 ; + cmd->raw_cmd[0 ] |= ((u64 )devid) <<32 ; +} + +static void its_encode_event_id (struct its_cmd_block *cmd,u32 id) +{ + cmd->raw_cmd[1 ] &= ~0xffffffffUL ; + cmd->raw_cmd[1 ] |= id; +} + +static void its_encode_phys_id (struct its_cmd_block *cmd,u32 phys_id) +{ + cmd->raw_cmd[1 ] &=0xffffffffUL ; + cmd->raw_cmd[1 ] |= ((u64 )phys_id) <<32 ; +} + +static void its_encode_size (struct its_cmd_block *cmd,u8 size) +{ + cmd->raw_cmd[1 ] &= ~0x1fUL ; + cmd->raw_cmd[1 ] |= size &0x1f ; +} + +static void its_encode_itt (struct its_cmd_block *cmd,u64 itt_addr) +{ + cmd->raw_cmd[2 ] &= ~0xffffffffffffUL ; + cmd->raw_cmd[2 ] |= itt_addr &0xffffffffff00UL ; +} + +static void its_encode_valid (struct its_cmd_block *cmd,int valid) +{ + cmd->raw_cmd[2 ] &= ~(1UL <<63 ); + cmd->raw_cmd[2 ] |= ((u64 )!!valid) <<63 ; +} + +static void its_encode_target (struct its_cmd_block *cmd,u64 target_addr) +{ + cmd->raw_cmd[2 ] &= ~(0xfffffffffUL <<16 ); + cmd->raw_cmd[2 ] |= (target_addr & (0xffffffffUL <<16 )); +} + +static void its_encode_collection (struct its_cmd_block *cmd,u16 col) +{ + cmd->raw_cmd[2 ] &= ~0xffffUL ; + cmd->raw_cmd[2 ] |= col; +} + +static inline void its_fixup_cmd (struct its_cmd_block *cmd) +{ +/* Let's fixup BE commands */ + cmd->raw_cmd[0 ] =cpu_to_le64 (cmd->raw_cmd[0 ]); + cmd->raw_cmd[1 ] =cpu_to_le64 (cmd->raw_cmd[1 ]); + cmd->raw_cmd[2 ] =cpu_to_le64 (cmd->raw_cmd[2 ]); + cmd->raw_cmd[3 ] =cpu_to_le64 (cmd->raw_cmd[3 ]); +} + +static u64 its_cmd_ptr_to_offset (struct its_cmd_block *ptr) +{ +return (ptr - its_data.cmd_base) *sizeof (*ptr); +}
lib/arm/gic-v3-its-cmd.c line 165
Quoted 10 lines
lib/arm/gic-v3-its-cmd.c new file +10
-0+ +static struct its_cmd_block *its_post_commands (void ) +{ +u64 wr =its_cmd_ptr_to_offset (its_data.cmd_write); + +writeq (wr, its_data.base +GITS_CWRITER ); +return its_data.cmd_write; +} + +
extra blank line
Quoted 20 lines
lib/arm/gic-v3-its-cmd.c new file +20
-0+static struct its_cmd_block *its_allocate_entry (void ) +{ +struct its_cmd_block *cmd; + +assert ((u64 )its_data.cmd_write < (u64 )its_data.cmd_base +SZ_64K ); + cmd = its_data.cmd_write++; +return cmd; +} + +static void its_wait_for_range_completion (struct its_cmd_block *from, +struct its_cmd_block *to) +{ +u64 rd_idx, from_idx, to_idx; +u32 count =1000000 ;/* 1s! */ + + from_idx =its_cmd_ptr_to_offset (from); + to_idx =its_cmd_ptr_to_offset (to); +while (1 ) { + rd_idx =readq (its_data.base +GITS_CREADR ); +if (rd_idx >= to_idx || rd_idx < from_idx)
lib/arm/gic-v3-its-cmd.c line 195
Quoted 10 lines
lib/arm/gic-v3-its-cmd.c new file +10
-0+break ; + + count--; +if (!count) { +unsigned int cmd_id = from->raw_cmd[0 ] &0xFF ; + +assert_msg (false,"%s timeout!" , + cmd_id <=0xF ? its_cmd_string[cmd_id] : +"Unexpected" ); +return ;
No need for 'return' after assert.
lib/arm/gic-v3-its-cmd.c line 197
Quoted 2 lines
lib/arm/gic-v3-its-cmd.c new file +2
-0+ } +cpu_relax ();
no need for cpu_relax right before udelay which calls cpu_relax
Quoted 42 lines
lib/arm/gic-v3-its-cmd.c new file +42
-0+udelay (1 ); + } +} + +static void its_send_single_command (its_cmd_builder_t builder, +struct its_cmd_desc *desc) +{ +struct its_cmd_block *cmd, *next_cmd; + + cmd =its_allocate_entry (); +builder (cmd, desc); + next_cmd =its_post_commands (); + +its_wait_for_range_completion (cmd, next_cmd); +} + + +static void its_build_mapd_cmd (struct its_cmd_block *cmd, +struct its_cmd_desc *desc) +{ +unsigned long itt_addr; +u8 size =12 ;/* 4096 eventids */ + + itt_addr = (unsigned long )desc->its_mapd_cmd.dev->itt; + itt_addr =ALIGN (itt_addr,ITS_ITT_ALIGN ); + +its_encode_cmd (cmd,GITS_CMD_MAPD ); +its_encode_devid (cmd, desc->its_mapd_cmd.dev->device_id); +its_encode_size (cmd, size -1 ); +its_encode_itt (cmd, itt_addr); +its_encode_valid (cmd, desc->its_mapd_cmd.valid); + +its_fixup_cmd (cmd); +printf ("ITS: MAPD devid=%d size = 0x%x itt=0x%lx valid=%d\n" , + desc->its_mapd_cmd.dev->device_id, + size, itt_addr, desc->its_mapd_cmd.valid); + +} + +static void its_build_mapc_cmd (struct its_cmd_block *cmd, +struct its_cmd_desc *desc) +{
lib/arm/gic-v3-its-cmd.c line 249
Quoted 10 lines
lib/arm/gic-v3-its-cmd.c new file +10
-0+its_encode_cmd (cmd,GITS_CMD_MAPC ); +its_encode_collection (cmd, desc->its_mapc_cmd.col->col_id); +its_encode_target (cmd, desc->its_mapc_cmd.col->target_address); +its_encode_valid (cmd, desc->its_mapc_cmd.valid); + +its_fixup_cmd (cmd); +report_info ("MAPC col_id=%d target_addr = 0x%lx valid=%d" , + desc->its_mapc_cmd.col->col_id, + desc->its_mapc_cmd.col->target_address, + desc->its_mapc_cmd.valid);
printf, but better yet, leave the printing to the callers. We're in library code here, so if a unit test doesn't want this verbosity then they shouldn't have to have it. Same comment for the above printf and all the below report_infos.
Quoted 73 lines
lib/arm/gic-v3-its-cmd.c new file +73
-0+} + +static void its_build_mapti_cmd (struct its_cmd_block *cmd, +struct its_cmd_desc *desc) +{ +its_encode_cmd (cmd,GITS_CMD_MAPTI ); +its_encode_devid (cmd, desc->its_mapti_cmd.dev->device_id); +its_encode_event_id (cmd, desc->its_mapti_cmd.event_id); +its_encode_phys_id (cmd, desc->its_mapti_cmd.phys_id); +its_encode_collection (cmd, desc->its_mapti_cmd.col_id); + +its_fixup_cmd (cmd); +report_info ("MAPTI dev_id=%d event_id=%d -> phys_id=%d, col_id=%d" , + desc->its_mapti_cmd.dev->device_id, + desc->its_mapti_cmd.event_id, + desc->its_mapti_cmd.phys_id, + desc->its_mapti_cmd.col_id); +} + +static void its_build_invall_cmd (struct its_cmd_block *cmd, +struct its_cmd_desc *desc) +{ +its_encode_cmd (cmd,GITS_CMD_INVALL ); +its_encode_collection (cmd, desc->its_invall_cmd.col->col_id); + +its_fixup_cmd (cmd); +report_info ("INVALL col_id=%d" , desc->its_invall_cmd.col->col_id); +} + +static void its_build_clear_cmd (struct its_cmd_block *cmd, +struct its_cmd_desc *desc) +{ +its_encode_cmd (cmd,GITS_CMD_CLEAR ); +its_encode_devid (cmd, desc->its_clear_cmd.dev->device_id); +its_encode_event_id (cmd, desc->its_clear_cmd.event_id); + +its_fixup_cmd (cmd); +report_info ("CLEAR col_id=%d" , desc->its_invall_cmd.col->col_id); +} + +static void its_build_discard_cmd (struct its_cmd_block *cmd, +struct its_cmd_desc *desc) +{ +its_encode_cmd (cmd,GITS_CMD_DISCARD ); +its_encode_devid (cmd, desc->its_discard_cmd.dev->device_id); +its_encode_event_id (cmd, desc->its_discard_cmd.event_id); + +its_fixup_cmd (cmd); +report_info ("DISCARD col_id=%d" , desc->its_invall_cmd.col->col_id); +} + +static void its_build_inv_cmd (struct its_cmd_block *cmd, +struct its_cmd_desc *desc) +{ +its_encode_cmd (cmd,GITS_CMD_INV ); +its_encode_devid (cmd, desc->its_inv_cmd.dev->device_id); +its_encode_event_id (cmd, desc->its_inv_cmd.event_id); + +its_fixup_cmd (cmd); +report_info ("INV dev_id=%d event_id=%d" , + desc->its_inv_cmd.dev->device_id, + desc->its_inv_cmd.event_id); +} + +static void its_build_int_cmd (struct its_cmd_block *cmd, +struct its_cmd_desc *desc) +{ +its_encode_cmd (cmd,GITS_CMD_INT ); +its_encode_devid (cmd, desc->its_int_cmd.dev->device_id); +its_encode_event_id (cmd, desc->its_int_cmd.event_id); + +its_fixup_cmd (cmd); + report_info("INT dev_id=%d event_id=%d",
lib/arm/gic-v3-its-cmd.c line 332
Quoted 10 lines
lib/arm/gic-v3-its-cmd.c new file +10
-0+ desc->its_int_cmd.dev->device_id, + desc->its_int_cmd.event_id); +} + +static void its_build_sync_cmd (struct its_cmd_block *cmd, +struct its_cmd_desc *desc) +{ +its_encode_cmd (cmd,GITS_CMD_SYNC ); +its_encode_target (cmd, desc->its_sync_cmd.col->target_address); +its_fixup_cmd (cmd);
All the rest of the blocks have a blank line before its_fixup_cmd, but I actually like this one better. The blanks are unnecessary.
Quoted 126 lines
lib/arm/gic-v3-its-cmd.c new file +122
-0+report_info ("SYNC target_addr = 0x%lx" , + desc->its_sync_cmd.col->target_address); +} + +static void its_build_movi_cmd (struct its_cmd_block *cmd, +struct its_cmd_desc *desc) +{ +its_encode_cmd (cmd,GITS_CMD_MOVI ); +its_encode_devid (cmd, desc->its_movi_cmd.dev->device_id); +its_encode_event_id (cmd, desc->its_movi_cmd.event_id); +its_encode_collection (cmd, desc->its_movi_cmd.col->col_id); + +its_fixup_cmd (cmd); +report_info ("MOVI dev_id=%d event_id = %d col_id=%d" , + desc->its_movi_cmd.dev->device_id, + desc->its_movi_cmd.event_id, + desc->its_movi_cmd.col->col_id); +} + +void its_send_mapd (struct its_device *dev,int valid) +{ +struct its_cmd_desc desc; + + desc.its_mapd_cmd.dev = dev; + desc.its_mapd_cmd.valid = !!valid; + +its_send_single_command (its_build_mapd_cmd, &desc); +} + +void its_send_mapc (struct its_collection *col,int valid) +{ +struct its_cmd_desc desc; + + desc.its_mapc_cmd.col = col; + desc.its_mapc_cmd.valid = !!valid; + +its_send_single_command (its_build_mapc_cmd, &desc); +} + +void its_send_mapti (struct its_device *dev,u32 irq_id, +u32 event_id,struct its_collection *col) +{ +struct its_cmd_desc desc; + + desc.its_mapti_cmd.dev = dev; + desc.its_mapti_cmd.phys_id = irq_id; + desc.its_mapti_cmd.event_id = event_id; + desc.its_mapti_cmd.col_id = col->col_id; + +its_send_single_command (its_build_mapti_cmd, &desc); +} + +void its_send_int (struct its_device *dev,u32 event_id) +{ +struct its_cmd_desc desc; + + desc.its_int_cmd.dev = dev; + desc.its_int_cmd.event_id = event_id; + +its_send_single_command (its_build_int_cmd, &desc); +} + +void its_send_movi (struct its_device *dev, +struct its_collection *col,u32 id) +{ +struct its_cmd_desc desc; + + desc.its_movi_cmd.dev = dev; + desc.its_movi_cmd.col = col; + desc.its_movi_cmd.event_id = id; + +its_send_single_command (its_build_movi_cmd, &desc); +} + +void its_send_invall (struct its_collection *col) +{ +struct its_cmd_desc desc; + + desc.its_invall_cmd.col = col; + +its_send_single_command (its_build_invall_cmd, &desc); +} + +void its_send_inv (struct its_device *dev,u32 event_id) +{ +struct its_cmd_desc desc; + + desc.its_inv_cmd.dev = dev; + desc.its_inv_cmd.event_id = event_id; + +its_send_single_command (its_build_inv_cmd, &desc); +} + +void its_send_discard (struct its_device *dev,u32 event_id) +{ +struct its_cmd_desc desc; + + desc.its_discard_cmd.dev = dev; + desc.its_discard_cmd.event_id = event_id; + +its_send_single_command (its_build_discard_cmd, &desc); +} + +void its_send_clear (struct its_device *dev,u32 event_id) +{ +struct its_cmd_desc desc; + + desc.its_clear_cmd.dev = dev; + desc.its_clear_cmd.event_id = event_id; + +its_send_single_command (its_build_clear_cmd, &desc); +} + +void its_send_sync (struct its_collection *col) +{ +struct its_cmd_desc desc; + + desc.its_sync_cmd.col = col; + +its_send_single_command (its_build_sync_cmd, &desc); +} +Signature · 1 line
2.20.1
Thanks,
drew
Re: [kvm-unit-tests PATCH v3 10/14] arm/arm64: ITS: commands
Hi Drew,
Quoted 42 lines · Andrew Jones
Quoted 41 lines · Eric Auger
Implement main ITS commands. The code is largely inherited from the ITS driver.
Signed-off-by Eric Auger <[email protected]>---
v2 -> v3:
- do not usereport()anymore
- assert ifcmd_writeexceeds the queue capacityv1 -> v2:
- removedits_print_cmd_state--- arm/Makefile.arm64 | 2 +- lib/arm/asm/gic-v3-its.h | 38 +++- lib/arm/gic-v3-its-cmd.c | 454 +++++++++++++++++++++++++++++++++++++++ 3 files changed, 492 insertions(+), 2 deletions(-) create mode 100644 lib/arm/gic-v3-its-cmd.carm/Makefile.arm64 +1
-1@@ -19,7 +19,7 @@ endef cstart.o = $(TEST_DIR)/cstart64.o cflatobjs += lib/arm64/processor.o cflatobjs += lib/arm64/spinlock.o -cflatobjs += lib/arm/gic-v3-its.o +cflatobjs += lib/arm/gic-v3-its.o lib/arm/gic-v3-its-cmd.o OBJDIRS += lib/arm64lib/arm/asm/gic-v3-its.h +0
-0@@ -45,6 +45,8 @@ struct its_collection {u16 col_id; };
Quoted 5 lines
lib/arm/asm/gic-v3-its.h struct its_collection { · line 49
Quoted 2 lines
lib/arm/asm/gic-v3-its.h +2
-0+struct its_cmd_block ; +This isn't necessary. If it was, then it should have been added in a previous patch.
removed
Quoted 19 lines
Quoted 19 lines
lib/arm/asm/gic-v3-its.h +11
-0struct its_data {void *base;struct its_typer typer;@@ -107,6 +109,24 @@ extern struct its_data its_data;#define GITS_BASER_TYPE_DEVICE 1#define GITS_BASER_TYPE_COLLECTION 4 +/* +* ITS commands +*/ +#define GITS_CMD_MAPD 0x08 +#define GITS_CMD_MAPC 0x09 +#define GITS_CMD_MAPTI 0x0a +/* older GIC documentation used MAPVI for this command */ +#define GITS_CMD_MAPVI GITS_CMD_MAPTI +#define GITS_CMD_MAPI 0x0b +#define GITS_CMD_MOVI 0x01 +#define GITS_CMD_DISCARD 0x0f
Quoted 8 lines
lib/arm/asm/gic-v3-its.h extern struct its_data its_data; · line 128
Quoted 6 lines
lib/arm/asm/gic-v3-its.h +6
-0+#define GITS_CMD_INV 0x0c +#define GITS_CMD_MOVALL 0x0e +#define GITS_CMD_INVALL 0x0d +#define GITS_CMD_INT 0x03 +#define GITS_CMD_CLEAR 0x04 +#define GITS_CMD_SYNC 0x05Please use tabs.
done
Quoted 19 lines
Quoted 19 lines
lib/arm/asm/gic-v3-its.h +11
-0+struct its_cmd_block {u64 raw_cmd[4 ]; };@@ -119,11 +139,27 @@ extern void its_enable_defaults(void);extern struct its_device *its_create_device (u32 dev_id,int nr_ites);extern struct its_collection *its_create_collection (u32 col_id,u32 target_pe); +extern void its_send_mapd (struct its_device *dev,int valid); +extern void its_send_mapc (struct its_collection *col,int valid); +extern void its_send_mapti (struct its_device *dev,u32 irq_id, +u32 event_id,struct its_collection *col); +extern void its_send_int (struct its_device *dev,u32 event_id); +extern void its_send_inv (struct its_device *dev,u32 event_id); +extern void its_send_discard (struct its_device *dev,u32 event_id); +extern void its_send_clear (struct its_device *dev,u32 event_id); +extern void its_send_invall (struct its_collection *col); +extern void its_send_movi(struct its_device *dev,
Quoted 8 lines
lib/arm/asm/gic-v3-its.h extern void its_enable_defaults(void); · line 157
Quoted 6 lines
lib/arm/asm/gic-v3-its.h +6
-0+ struct its_collection *col, u32 id); +extern void its_send_sync (struct its_collection *col); + +#define ITS_FLAGS_CMDQ_NEEDS_FLUSHING (1ULL << 0) +#define ITS_FLAGS_WORKAROUND_CAVIUM_22375 (1ULL << 1) +#define ITS_FLAGS_WORKAROUND_CAVIUM_23144 (1ULL << 2)What are these flags for?
removed
Quoted 22 lines
Quoted 22 lines
lib/arm/asm/gic-v3-its.h +1
-1+#else /* __arm__ */ static inline void its_init (void ) {}#endif -#endif /* !__ASSEMBLY__ */ #endif /* _ASMARM_GIC_V3_ITS_H_ */ lib/arm/gic-v3-its-cmd.c new file +6
-0@@ -0,0 +1,454 @@ +/* +* Copyright (C) 2020, Red Hat Inc, Eric Auger <[email protected]> +* +* Most of the code is copy-pasted from: +* drivers/irqchip/irq-gic-v3-its.c +* This work is licensed under the terms of the GNU LGPL, version 2.
Quoted 8 lines
lib/arm/gic-v3-its-cmd.c line 12
Quoted 6 lines
lib/arm/gic-v3-its-cmd.c new file +6
-0+*/ +#include <asm/io.h> +#include <asm/gic.h> +#include <asm/gic-v3-its.h> + +#define ITS_ITT_ALIGN SZ_256tabs
done
Quoted 148 lines
Quoted 148 lines
lib/arm/gic-v3-its-cmd.c new file +147
-0+ +static const char *const its_cmd_string[] = { + [GITS_CMD_MAPD ] ="MAPD" , + [GITS_CMD_MAPC ] ="MAPC" , + [GITS_CMD_MAPTI ] ="MAPTI" , + [GITS_CMD_MAPI ] ="MAPI" , + [GITS_CMD_MOVI ] ="MOVI" , + [GITS_CMD_DISCARD ] ="DISCARD" , + [GITS_CMD_INV ] ="INV" , + [GITS_CMD_MOVALL ] ="MOVALL" , + [GITS_CMD_INVALL ] ="INVALL" , + [GITS_CMD_INT ] ="INT" , + [GITS_CMD_CLEAR ] ="CLEAR" , + [GITS_CMD_SYNC ] ="SYNC" , +}; + +struct its_cmd_desc { +union { +struct { +struct its_device *dev; +u32 event_id; + } its_inv_cmd; + +struct { +struct its_device *dev; +u32 event_id; + } its_int_cmd; + +struct { +struct its_device *dev; +bool valid; + } its_mapd_cmd; + +struct { +struct its_collection *col; +bool valid; + } its_mapc_cmd; + +struct { +struct its_device *dev; +u32 phys_id; +u32 event_id; +u32 col_id; + } its_mapti_cmd; + +struct { +struct its_device *dev; +struct its_collection *col; +u32 event_id; + } its_movi_cmd; + +struct { +struct its_device *dev; +u32 event_id; + } its_discard_cmd; + +struct { +struct its_device *dev; +u32 event_id; + } its_clear_cmd; + +struct { +struct its_collection *col; + } its_invall_cmd; + +struct { +struct its_collection *col; + } its_sync_cmd; + }; +}; + +typedef void (*its_cmd_builder_t )(struct its_cmd_block *, +struct its_cmd_desc *); + +/* ITS COMMANDS */ + +static void its_encode_cmd (struct its_cmd_block *cmd,u8 cmd_nr) +{ + cmd->raw_cmd[0 ] &= ~0xffUL ; + cmd->raw_cmd[0 ] |= cmd_nr; +} + +static void its_encode_devid (struct its_cmd_block *cmd,u32 devid) +{ + cmd->raw_cmd[0 ] &=BIT_ULL (32 ) -1 ; + cmd->raw_cmd[0 ] |= ((u64 )devid) <<32 ; +} + +static void its_encode_event_id (struct its_cmd_block *cmd,u32 id) +{ + cmd->raw_cmd[1 ] &= ~0xffffffffUL ; + cmd->raw_cmd[1 ] |= id; +} + +static void its_encode_phys_id (struct its_cmd_block *cmd,u32 phys_id) +{ + cmd->raw_cmd[1 ] &=0xffffffffUL ; + cmd->raw_cmd[1 ] |= ((u64 )phys_id) <<32 ; +} + +static void its_encode_size (struct its_cmd_block *cmd,u8 size) +{ + cmd->raw_cmd[1 ] &= ~0x1fUL ; + cmd->raw_cmd[1 ] |= size &0x1f ; +} + +static void its_encode_itt (struct its_cmd_block *cmd,u64 itt_addr) +{ + cmd->raw_cmd[2 ] &= ~0xffffffffffffUL ; + cmd->raw_cmd[2 ] |= itt_addr &0xffffffffff00UL ; +} + +static void its_encode_valid (struct its_cmd_block *cmd,int valid) +{ + cmd->raw_cmd[2 ] &= ~(1UL <<63 ); + cmd->raw_cmd[2 ] |= ((u64 )!!valid) <<63 ; +} + +static void its_encode_target (struct its_cmd_block *cmd,u64 target_addr) +{ + cmd->raw_cmd[2 ] &= ~(0xfffffffffUL <<16 ); + cmd->raw_cmd[2 ] |= (target_addr & (0xffffffffUL <<16 )); +} + +static void its_encode_collection (struct its_cmd_block *cmd,u16 col) +{ + cmd->raw_cmd[2 ] &= ~0xffffUL ; + cmd->raw_cmd[2 ] |= col; +} + +static inline void its_fixup_cmd (struct its_cmd_block *cmd) +{ +/* Let's fixup BE commands */ + cmd->raw_cmd[0 ] =cpu_to_le64 (cmd->raw_cmd[0 ]); + cmd->raw_cmd[1 ] =cpu_to_le64 (cmd->raw_cmd[1 ]); + cmd->raw_cmd[2 ] =cpu_to_le64 (cmd->raw_cmd[2 ]); + cmd->raw_cmd[3 ] =cpu_to_le64 (cmd->raw_cmd[3 ]); +} + +static u64 its_cmd_ptr_to_offset (struct its_cmd_block *ptr) +{ +return (ptr - its_data.cmd_base) *sizeof (*ptr); +} + +static struct its_cmd_block *its_post_commands (void ) +{ +u64 wr =its_cmd_ptr_to_offset (its_data.cmd_write);
Quoted 8 lines
lib/arm/gic-v3-its-cmd.c line 165
Quoted 6 lines
lib/arm/gic-v3-its-cmd.c new file +6
-0+ +writeq (wr, its_data.base +GITS_CWRITER ); +return its_data.cmd_write; +} + +extra blank line
removed
Quoted 25 lines
Quoted 25 lines
lib/arm/gic-v3-its-cmd.c new file +24
-0+static struct its_cmd_block *its_allocate_entry (void ) +{ +struct its_cmd_block *cmd; + +assert ((u64 )its_data.cmd_write < (u64 )its_data.cmd_base +SZ_64K ); + cmd = its_data.cmd_write++; +return cmd; +} + +static void its_wait_for_range_completion (struct its_cmd_block *from, +struct its_cmd_block *to) +{ +u64 rd_idx, from_idx, to_idx; +u32 count =1000000 ;/* 1s! */ + + from_idx =its_cmd_ptr_to_offset (from); + to_idx =its_cmd_ptr_to_offset (to); +while (1 ) { + rd_idx =readq (its_data.base +GITS_CREADR ); +if (rd_idx >= to_idx || rd_idx < from_idx) +break ; + + count--; +if (!count) {
Quoted 8 lines
lib/arm/gic-v3-its-cmd.c line 195
Quoted 6 lines
lib/arm/gic-v3-its-cmd.c new file +6
-0+unsigned int cmd_id = from->raw_cmd[0 ] &0xFF ; + +assert_msg (false,"%s timeout!" , + cmd_id <=0xF ? its_cmd_string[cmd_id] : +"Unexpected" ); +return ;No need for 'return' after assert.
removed
Quoted 5 lines
lib/arm/gic-v3-its-cmd.c line 197
Quoted 3 lines
lib/arm/gic-v3-its-cmd.c new file +2
-0+ } +cpu_relax ();no need for
cpu_relaxright before udelay which callscpu_relax
removed
Quoted 50 lines
Quoted 50 lines
lib/arm/gic-v3-its-cmd.c new file +49
-0+udelay (1 ); + } +} + +static void its_send_single_command (its_cmd_builder_t builder, +struct its_cmd_desc *desc) +{ +struct its_cmd_block *cmd, *next_cmd; + + cmd =its_allocate_entry (); +builder (cmd, desc); + next_cmd =its_post_commands (); + +its_wait_for_range_completion (cmd, next_cmd); +} + + +static void its_build_mapd_cmd (struct its_cmd_block *cmd, +struct its_cmd_desc *desc) +{ +unsigned long itt_addr; +u8 size =12 ;/* 4096 eventids */ + + itt_addr = (unsigned long )desc->its_mapd_cmd.dev->itt; + itt_addr =ALIGN (itt_addr,ITS_ITT_ALIGN ); + +its_encode_cmd (cmd,GITS_CMD_MAPD ); +its_encode_devid (cmd, desc->its_mapd_cmd.dev->device_id); +its_encode_size (cmd, size -1 ); +its_encode_itt (cmd, itt_addr); +its_encode_valid (cmd, desc->its_mapd_cmd.valid); + +its_fixup_cmd (cmd); +printf ("ITS: MAPD devid=%d size = 0x%x itt=0x%lx valid=%d\n" , + desc->its_mapd_cmd.dev->device_id, + size, itt_addr, desc->its_mapd_cmd.valid); + +} + +static void its_build_mapc_cmd (struct its_cmd_block *cmd, +struct its_cmd_desc *desc) +{ +its_encode_cmd (cmd,GITS_CMD_MAPC ); +its_encode_collection (cmd, desc->its_mapc_cmd.col->col_id); +its_encode_target (cmd, desc->its_mapc_cmd.col->target_address); +its_encode_valid (cmd, desc->its_mapc_cmd.valid); + +its_fixup_cmd (cmd); + report_info("MAPC col_id=%d target_addr = 0x%lx valid=%d",
Quoted 8 lines
lib/arm/gic-v3-its-cmd.c line 249
Quoted 3 lines
lib/arm/gic-v3-its-cmd.c new file +3
-0+ desc->its_mapc_cmd.col->col_id, + desc->its_mapc_cmd.col->target_address, + desc->its_mapc_cmd.valid);printf, but better yet, leave the printing to the callers. We're in library code here, so if a unit test doesn't want this verbosity then they shouldn't have to have it. Same comment for the above printf and all the below
report_infos.
I switched to printf. I added a verbose field to desc so that send* function can pass a verbose argument. It is easier to trace at low levels as all the fields are resolved there.
Quoted 79 lines
Quoted 79 lines
lib/arm/gic-v3-its-cmd.c new file +78
-0+} + +static void its_build_mapti_cmd (struct its_cmd_block *cmd, +struct its_cmd_desc *desc) +{ +its_encode_cmd (cmd,GITS_CMD_MAPTI ); +its_encode_devid (cmd, desc->its_mapti_cmd.dev->device_id); +its_encode_event_id (cmd, desc->its_mapti_cmd.event_id); +its_encode_phys_id (cmd, desc->its_mapti_cmd.phys_id); +its_encode_collection (cmd, desc->its_mapti_cmd.col_id); + +its_fixup_cmd (cmd); +report_info ("MAPTI dev_id=%d event_id=%d -> phys_id=%d, col_id=%d" , + desc->its_mapti_cmd.dev->device_id, + desc->its_mapti_cmd.event_id, + desc->its_mapti_cmd.phys_id, + desc->its_mapti_cmd.col_id); +} + +static void its_build_invall_cmd (struct its_cmd_block *cmd, +struct its_cmd_desc *desc) +{ +its_encode_cmd (cmd,GITS_CMD_INVALL ); +its_encode_collection (cmd, desc->its_invall_cmd.col->col_id); + +its_fixup_cmd (cmd); +report_info ("INVALL col_id=%d" , desc->its_invall_cmd.col->col_id); +} + +static void its_build_clear_cmd (struct its_cmd_block *cmd, +struct its_cmd_desc *desc) +{ +its_encode_cmd (cmd,GITS_CMD_CLEAR ); +its_encode_devid (cmd, desc->its_clear_cmd.dev->device_id); +its_encode_event_id (cmd, desc->its_clear_cmd.event_id); + +its_fixup_cmd (cmd); +report_info ("CLEAR col_id=%d" , desc->its_invall_cmd.col->col_id); +} + +static void its_build_discard_cmd (struct its_cmd_block *cmd, +struct its_cmd_desc *desc) +{ +its_encode_cmd (cmd,GITS_CMD_DISCARD ); +its_encode_devid (cmd, desc->its_discard_cmd.dev->device_id); +its_encode_event_id (cmd, desc->its_discard_cmd.event_id); + +its_fixup_cmd (cmd); +report_info ("DISCARD col_id=%d" , desc->its_invall_cmd.col->col_id); +} + +static void its_build_inv_cmd (struct its_cmd_block *cmd, +struct its_cmd_desc *desc) +{ +its_encode_cmd (cmd,GITS_CMD_INV ); +its_encode_devid (cmd, desc->its_inv_cmd.dev->device_id); +its_encode_event_id (cmd, desc->its_inv_cmd.event_id); + +its_fixup_cmd (cmd); +report_info ("INV dev_id=%d event_id=%d" , + desc->its_inv_cmd.dev->device_id, + desc->its_inv_cmd.event_id); +} + +static void its_build_int_cmd (struct its_cmd_block *cmd, +struct its_cmd_desc *desc) +{ +its_encode_cmd (cmd,GITS_CMD_INT ); +its_encode_devid (cmd, desc->its_int_cmd.dev->device_id); +its_encode_event_id (cmd, desc->its_int_cmd.event_id); + +its_fixup_cmd (cmd); +report_info ("INT dev_id=%d event_id=%d" , + desc->its_int_cmd.dev->device_id, + desc->its_int_cmd.event_id); +} + +static void its_build_sync_cmd(struct its_cmd_block *cmd,
Quoted 8 lines
lib/arm/gic-v3-its-cmd.c line 332
Quoted 5 lines
lib/arm/gic-v3-its-cmd.c new file +5
-0+ struct its_cmd_desc *desc) +{ +its_encode_cmd (cmd,GITS_CMD_SYNC ); +its_encode_target (cmd, desc->its_sync_cmd.col->target_address); +its_fixup_cmd (cmd);All the rest of the blocks have a blank line before
its_fixup_cmd, but I actually like this one better. The blanks are unnecessary.
removed
Quoted 122 lines
Quoted 122 lines
lib/arm/gic-v3-its-cmd.c new file +121
-0+report_info ("SYNC target_addr = 0x%lx" , + desc->its_sync_cmd.col->target_address); +} + +static void its_build_movi_cmd (struct its_cmd_block *cmd, +struct its_cmd_desc *desc) +{ +its_encode_cmd (cmd,GITS_CMD_MOVI ); +its_encode_devid (cmd, desc->its_movi_cmd.dev->device_id); +its_encode_event_id (cmd, desc->its_movi_cmd.event_id); +its_encode_collection (cmd, desc->its_movi_cmd.col->col_id); + +its_fixup_cmd (cmd); +report_info ("MOVI dev_id=%d event_id = %d col_id=%d" , + desc->its_movi_cmd.dev->device_id, + desc->its_movi_cmd.event_id, + desc->its_movi_cmd.col->col_id); +} + +void its_send_mapd (struct its_device *dev,int valid) +{ +struct its_cmd_desc desc; + + desc.its_mapd_cmd.dev = dev; + desc.its_mapd_cmd.valid = !!valid; + +its_send_single_command (its_build_mapd_cmd, &desc); +} + +void its_send_mapc (struct its_collection *col,int valid) +{ +struct its_cmd_desc desc; + + desc.its_mapc_cmd.col = col; + desc.its_mapc_cmd.valid = !!valid; + +its_send_single_command (its_build_mapc_cmd, &desc); +} + +void its_send_mapti (struct its_device *dev,u32 irq_id, +u32 event_id,struct its_collection *col) +{ +struct its_cmd_desc desc; + + desc.its_mapti_cmd.dev = dev; + desc.its_mapti_cmd.phys_id = irq_id; + desc.its_mapti_cmd.event_id = event_id; + desc.its_mapti_cmd.col_id = col->col_id; + +its_send_single_command (its_build_mapti_cmd, &desc); +} + +void its_send_int (struct its_device *dev,u32 event_id) +{ +struct its_cmd_desc desc; + + desc.its_int_cmd.dev = dev; + desc.its_int_cmd.event_id = event_id; + +its_send_single_command (its_build_int_cmd, &desc); +} + +void its_send_movi (struct its_device *dev, +struct its_collection *col,u32 id) +{ +struct its_cmd_desc desc; + + desc.its_movi_cmd.dev = dev; + desc.its_movi_cmd.col = col; + desc.its_movi_cmd.event_id = id; + +its_send_single_command (its_build_movi_cmd, &desc); +} + +void its_send_invall (struct its_collection *col) +{ +struct its_cmd_desc desc; + + desc.its_invall_cmd.col = col; + +its_send_single_command (its_build_invall_cmd, &desc); +} + +void its_send_inv (struct its_device *dev,u32 event_id) +{ +struct its_cmd_desc desc; + + desc.its_inv_cmd.dev = dev; + desc.its_inv_cmd.event_id = event_id; + +its_send_single_command (its_build_inv_cmd, &desc); +} + +void its_send_discard (struct its_device *dev,u32 event_id) +{ +struct its_cmd_desc desc; + + desc.its_discard_cmd.dev = dev; + desc.its_discard_cmd.event_id = event_id; + +its_send_single_command (its_build_discard_cmd, &desc); +} + +void its_send_clear (struct its_device *dev,u32 event_id) +{ +struct its_cmd_desc desc; + + desc.its_clear_cmd.dev = dev; + desc.its_clear_cmd.event_id = event_id; + +its_send_single_command (its_build_clear_cmd, &desc); +} + +void its_send_sync (struct its_collection *col) +{ +struct its_cmd_desc desc; + + desc.its_sync_cmd.col = col; + +its_send_single_command (its_build_sync_cmd, &desc); +}
Quoted 9 lines
Quoted 5 lines
lib/arm/gic-v3-its-cmd.c new file +1
-0+Signature · 1 line
2.20.1Thanks,
drew
Thanks
Eric
PATCH v3 11/14 arm/arm64: ITS: INT functional tests
In reply to Eric AugerTriggers LPIs through the INT command.
the test checks the LPI hits the right CPU and triggers the right LPI intid, ie. the translation is correct.
Updates to the config table also are tested, along with inv and invall commands.
---
v2 -> v3:
- add comments
- keep the report_skip in case there aren't 4 vcpus to be able to run other tests in the its category.
- fix the prefix pop
- move its_event and its_stats to arm/gic.c
--- arm/gic.c | 228 +++++++++++++++++++++++++++++++++++++++++++--- arm/unittests.cfg | 7 ++ 2 files changed, 224 insertions(+), 11 deletions(-)
arm/gic.c +217 -11
@@ -160,6 +160,87 @@ static void ipi_handler(struct pt_regs *regs __unused) } } +static void setup_irq (handler_t handler) +{ +gic_enable_defaults (); +#ifdef __arm__ +install_exception_handler (EXCPTN_IRQ , handler); +#else +install_irq_handler (EL1H_IRQ , handler); +#endif +local_irq_enable (); +} + +#if defined(__aarch64__) +struct its_event { +int cpu_id; +int lpi_id; +}; + +struct its_stats { +struct its_event expected; +struct its_event observed; +}; + +static struct its_stats lpi_stats; + +static void lpi_handler(struct pt_regs *regs __unused) +{ +u32 irqstat =gic_read_iar (); +int irqnr =gic_iar_irqnr (irqstat); + +gic_write_eoir (irqstat); +if (irqnr <8192 ) +report (false,"Unexpected non LPI interrupt received" ); +smp_rmb ();/* pairs with wmb in lpi_stats_expect */ + lpi_stats.observed.cpu_id =smp_processor_id (); + lpi_stats.observed.lpi_id = irqnr; +smp_wmb ();/* pairs with rmb in check_lpi_stats */ +} + +static void lpi_stats_expect (int exp_cpu_id,int exp_lpi_id) +{ + lpi_stats.expected.cpu_id = exp_cpu_id; + lpi_stats.expected.lpi_id = exp_lpi_id; + lpi_stats.observed.cpu_id =-1 ; + lpi_stats.observed.lpi_id =-1 ; +smp_wmb ();/* pairs with rmb in handler */ +} + +static void check_lpi_stats (void ) +{ +mdelay (100 ); +smp_rmb ();/* pairs with wmb in lpi_handler */ +if ((lpi_stats.observed.cpu_id != lpi_stats.expected.cpu_id) || + (lpi_stats.observed.lpi_id != lpi_stats.expected.lpi_id)) { +if (lpi_stats.observed.cpu_id ==-1 && + lpi_stats.observed.lpi_id ==-1 ) { +report (false, +"No LPI received whereas (cpuid=%d, intid=%d) " +"was expected" , lpi_stats.expected.cpu_id, + lpi_stats.expected.lpi_id); + }else { +report (false,"Unexpected LPI (cpuid=%d, intid=%d)" , + lpi_stats.observed.cpu_id, + lpi_stats.observed.lpi_id); + } + }else if (lpi_stats.expected.lpi_id !=-1 ) { +report (true,"LPI %d on CPU %d" , lpi_stats.observed.lpi_id, + lpi_stats.observed.cpu_id); + }else { +report (true,"no LPI received, as expected" ); + } +} + +static void secondary_lpi_test (void ) +{ +setup_irq (lpi_handler); +cpumask_set_cpu (smp_processor_id (), &ready); +while (1 ) +wfi (); +} +#endif +static void gicv2_ipi_send_self (void ) {writel (2 <<24 |IPI_IRQ ,gicv2_dist_base () +GICD_SGIR );
@@ -217,17 +298,6 @@ static void ipi_test_smp(void)report_prefix_pop (); } -static void setup_irq (handler_t handler) -{ -gic_enable_defaults (); -#ifdef __arm__ -install_exception_handler (EXCPTN_IRQ , handler); -#else -install_irq_handler (EL1H_IRQ , handler); -#endif -local_irq_enable (); -} -static void ipi_send (void ) {setup_irq (ipi_handler);
@@ -522,6 +592,7 @@ static void gic_test_mmio(void)#if defined(__arm__)static void test_its_introspection (void ) {} +static void test_its_trigger (void ) {}#else /* __arch64__ */
@@ -561,6 +632,137 @@ static void test_its_introspection(void)report_info ("collection baser entry_size = 0x%x" , coll_baser->esz); } +static bool its_prerequisites (int nb_cpus) +{ +int cpu; + +if (!gicv3_its_base ()) { +report_skip ("No ITS, skip ..." ); +return true; + } + +if (nr_cpus <4 ) { +report_skip ("Test requires at least %d vcpus" , nb_cpus); +return true; + } + +stats_reset (); + +setup_irq (lpi_handler); + +for_each_present_cpu (cpu) { +if (cpu ==0 ) +continue ; +smp_boot_secondary (cpu, secondary_lpi_test); + } +wait_on_ready (); + +its_enable_defaults (); + +lpi_stats_expect (-1 ,-1 ); +check_lpi_stats (); + +return false; +} + +static void test_its_trigger (void ) +{ +struct its_collection *col3, *col2; +struct its_device *dev2, *dev7; + +if (its_prerequisites (4 )) +return ; + + dev2 =its_create_device (2 /* dev id */ ,8 /* nb_ites */ ); + dev7 =its_create_device (7 /* dev id */ ,8 /* nb_ites */ ); + + col3 =its_create_collection (3 /* col id */ ,3 /* target PE */ ); + col2 =its_create_collection (2 /* col id */ ,2 /* target PE */ ); + +gicv3_lpi_set_config (8195 ,LPI_PROP_DEFAULT ); +gicv3_lpi_set_config (8196 ,LPI_PROP_DEFAULT ); + +its_send_invall (col2); +its_send_invall (col3); + +report_prefix_push ("int" ); +/* +* dev=2, eventid=20 -> lpi= 8195, col=3 +* dev=7, eventid=255 -> lpi= 8196, col=2 +* Trigger dev2, eventid=20 and dev7, eventid=255 +* Check both LPIs hit +*/ + +its_send_mapd (dev2, true); +its_send_mapd (dev7, true); + +its_send_mapc (col3, true); +its_send_mapc (col2, true); + +its_send_mapti (dev2,8195 /* lpi id */ , +20 /* event id */ , col3); +its_send_mapti (dev7,8196 /* lpi id */ , +255 /* event id */ , col2); + +lpi_stats_expect (3 ,8195 ); +its_send_int (dev2,20 ); +check_lpi_stats (); + +lpi_stats_expect (2 ,8196 ); +its_send_int (dev7,255 ); +check_lpi_stats (); + +report_prefix_pop (); + +report_prefix_push ("inv/invall" ); + +/* +* disable 8195, check dev2/eventid=20 does not trigger the +* corresponding LPI +*/ +gicv3_lpi_set_config (8195 ,LPI_PROP_DEFAULT & ~0x1 ); +its_send_inv (dev2,20 ); + +lpi_stats_expect (-1 ,-1 ); +its_send_int (dev2,20 ); +check_lpi_stats (); + +/* +* re-enable the LPI but willingly do not call invall +* so the change in config is not taken into account. +* The LPI should not hit +*/ +gicv3_lpi_set_config (8195 ,LPI_PROP_DEFAULT ); +lpi_stats_expect (-1 ,-1 ); +its_send_int (dev2,20 ); +check_lpi_stats (); + +/* Now call the invall and check the LPI hits */ +its_send_invall (col3); +lpi_stats_expect (3 ,8195 ); +its_send_int (dev2,20 ); +check_lpi_stats (); + +report_prefix_pop (); +/* +* Unmap device 2 and check the eventid 20 formerly +* attached to it does not hit anymore +*/ +report_prefix_push ("mapd valid=false" ); +its_send_mapd (dev2, false); +lpi_stats_expect (-1 ,-1 ); +its_send_int (dev2,20 ); +check_lpi_stats (); +report_prefix_pop (); + +/* Unmap the collection this time and check no LPI does hit */ +report_prefix_push ("mapc valid=false" ); +its_send_mapc (col2, false); +lpi_stats_expect (-1 ,-1 ); +its_send_int (dev7,255 ); +check_lpi_stats (); +report_prefix_pop (); +}#endif int main (int argc,char **argv)
@@ -594,6 +796,10 @@ int main(int argc, char **argv)report_prefix_push (argv[1 ]);gic_test_mmio ();report_prefix_pop (); + }else if (!strcmp (argv[1 ],"its-trigger" )) { +report_prefix_push (argv[1 ]); +test_its_trigger (); +report_prefix_pop (); }else if (strcmp (argv[1 ],"its-introspection" ) ==0 ) {report_prefix_push (argv[1 ]);test_its_introspection ();
arm/unittests.cfg +7 -0
@@ -129,6 +129,13 @@ extra_params = -machine gic-version=3 -append 'its-introspection' groups = its arch = arm64 +[its-trigger] +file = gic.flat +smp = $MAX_SMP +extra_params = -machine gic-version=3 -append 'its-trigger' +groups = its +arch = arm64 + # Test PSCI emulation [psci] file = psci.flat
Signature · 1 line
2.20.1
Re: [kvm-unit-tests PATCH v3 11/14] arm/arm64: ITS: INT functional tests
Quoted 53 lines · Eric Auger
Triggers LPIs through the INT command.
the test checks the LPI hits the right CPU and triggers the right LPI intid, ie. the translation is correct.
Updates to the config table also are tested, along with inv and invall commands.
Signed-off-by Eric Auger <[email protected]>---
v2 -> v3:
- add comments
- keep thereport_skipin case there aren't 4 vcpus to be able to run other tests in the its category.
- fix the prefix pop
- moveits_eventandits_statstoarm/gic.c--- arm/gic.c | 228 +++++++++++++++++++++++++++++++++++++++++++--- arm/unittests.cfg | 7 ++ 2 files changed, 224 insertions(+), 11 deletions(-)arm/gic.c +22
-0@@ -160,6 +160,87 @@ static void ipi_handler(struct pt_regs *regs __unused) } } +static void setup_irq (handler_t handler) +{ +gic_enable_defaults (); +#ifdef __arm__ +install_exception_handler (EXCPTN_IRQ , handler); +#else +install_irq_handler (EL1H_IRQ , handler); +#endif +local_irq_enable (); +} + +#if defined(__aarch64__) +struct its_event { +int cpu_id; +int lpi_id; +}; + +struct its_stats { +struct its_event expected; +struct its_event observed; +}; +
arm/gic.c static void ipi_handler(struct pt_regs *regs __unused) · line 194
Quoted 10 lines
arm/gic.c +10
-0+static struct its_stats lpi_stats; + +static void lpi_handler(struct pt_regs *regs __unused) +{ +u32 irqstat =gic_read_iar (); +int irqnr =gic_iar_irqnr (irqstat); + +gic_write_eoir (irqstat); +if (irqnr <8192 ) +report (false,"Unexpected non LPI interrupt received" );
report_info
Quoted 6 lines
arm/gic.c +6
-0+smp_rmb ();/* pairs with wmb in lpi_stats_expect */ + lpi_stats.observed.cpu_id =smp_processor_id (); + lpi_stats.observed.lpi_id = irqnr; +smp_wmb ();/* pairs with rmb in check_lpi_stats */ +} +
arm/gic.c static void ipi_handler(struct pt_regs *regs __unused) · line 210
Quoted 10 lines
arm/gic.c +10
-0+static void lpi_stats_expect (int exp_cpu_id,int exp_lpi_id) +{ + lpi_stats.expected.cpu_id = exp_cpu_id; + lpi_stats.expected.lpi_id = exp_lpi_id; + lpi_stats.observed.cpu_id =-1 ; + lpi_stats.observed.lpi_id =-1 ; +smp_wmb ();/* pairs with rmb in handler */ +} + +static void check_lpi_stats (void )
static void check_lpi_stats(const char *testname)
{
bool pass = false;
arm/gic.c static void ipi_handler(struct pt_regs *regs __unused) · line 215
Quoted 5 lines
arm/gic.c +5
-0+{ +mdelay (100 ); +smp_rmb ();/* pairs with wmb in lpi_handler */ +if ((lpi_stats.observed.cpu_id != lpi_stats.expected.cpu_id) || + (lpi_stats.observed.lpi_id != lpi_stats.expected.lpi_id)) {
nit: extra ()
arm/gic.c static void ipi_handler(struct pt_regs *regs __unused) · line 221
Quoted 6 lines
arm/gic.c +6
-0+if (lpi_stats.observed.cpu_id ==-1 && + lpi_stats.observed.lpi_id ==-1 ) { +report (false, +"No LPI received whereas (cpuid=%d, intid=%d) " +"was expected" , lpi_stats.expected.cpu_id, + lpi_stats.expected.lpi_id);
report_info
arm/gic.c static void ipi_handler(struct pt_regs *regs __unused) · line 225
Quoted 4 lines
arm/gic.c +4
-0+ }else { +report (false,"Unexpected LPI (cpuid=%d, intid=%d)" , + lpi_stats.observed.cpu_id, + lpi_stats.observed.lpi_id);
report_info
arm/gic.c static void ipi_handler(struct pt_regs *regs __unused) · line 226
Quoted 1 line
arm/gic.c +1
-0+ }
pass = false;
arm/gic.c static void ipi_handler(struct pt_regs *regs __unused) · line 229
Quoted 3 lines
arm/gic.c +3
-0+ }else if (lpi_stats.expected.lpi_id !=-1 ) { +report (true,"LPI %d on CPU %d" , lpi_stats.observed.lpi_id, + lpi_stats.observed.cpu_id);
report_info
arm/gic.c static void ipi_handler(struct pt_regs *regs __unused) · line 231
Quoted 2 lines
arm/gic.c +2
-0+ }else { +report (true,"no LPI received, as expected" );
report_info
arm/gic.c static void ipi_handler(struct pt_regs *regs __unused) · line 232
Quoted 1 line
arm/gic.c +1
-0+ }
report(pass, "%s", testname);
Quoted 44 lines
arm/gic.c +12
-11+} + +static void secondary_lpi_test (void ) +{ +setup_irq (lpi_handler); +cpumask_set_cpu (smp_processor_id (), &ready); +while (1 ) +wfi (); +} +#endif +static void gicv2_ipi_send_self (void ) {writel (2 <<24 |IPI_IRQ ,gicv2_dist_base () +GICD_SGIR );@@ -217,17 +298,6 @@ static void ipi_test_smp(void)report_prefix_pop (); } -static void setup_irq (handler_t handler) -{ -gic_enable_defaults (); -#ifdef __arm__ -install_exception_handler (EXCPTN_IRQ , handler); -#else -install_irq_handler (EL1H_IRQ , handler); -#endif -local_irq_enable (); -} -static void ipi_send (void ) {setup_irq (ipi_handler);@@ -522,6 +592,7 @@ static void gic_test_mmio(void)#if defined(__arm__)static void test_its_introspection (void ) {} +static void test_its_trigger (void ) {}#else /* __arch64__ */ @@ -561,6 +632,137 @@ static void test_its_introspection(void)report_info ("collection baser entry_size = 0x%x" , coll_baser->esz); }
arm/gic.c static void test_its_introspection(void) · line 644
Quoted 10 lines
arm/gic.c +10
-0+static bool its_prerequisites (int nb_cpus) +{ +int cpu; + +if (!gicv3_its_base ()) { +report_skip ("No ITS, skip ..." ); +return true; + } + +if (nr_cpus <4 ) {
nr_cpus < nb_cpus, or just drop the nb_cpus parameter and hard code 4 here.
Quoted 11 lines
arm/gic.c +11
-0+report_skip ("Test requires at least %d vcpus" , nb_cpus); +return true; + } + +stats_reset (); + +setup_irq (lpi_handler); + +for_each_present_cpu (cpu) { +if (cpu ==0 ) +continue ;
arm/gic.c static void test_its_introspection(void) · line 665
Quoted 10 lines
arm/gic.c +10
-0+smp_boot_secondary (cpu, secondary_lpi_test); + } +wait_on_ready (); + +its_enable_defaults (); + +lpi_stats_expect (-1 ,-1 ); +check_lpi_stats (); + +return false;
Reverse logic. I'd expect 'return true' for success.
arm/gic.c static void test_its_introspection(void) · line 673
Quoted 8 lines
arm/gic.c +8
-0+} + +static void test_its_trigger (void ) +{ +struct its_collection *col3, *col2; +struct its_device *dev2, *dev7; + +if (its_prerequisites (4 ))
if (!its_prerequisites(...))
Quoted 22 lines
arm/gic.c +22
-0+return ; + + dev2 =its_create_device (2 /* dev id */ ,8 /* nb_ites */ ); + dev7 =its_create_device (7 /* dev id */ ,8 /* nb_ites */ ); + + col3 =its_create_collection (3 /* col id */ ,3 /* target PE */ ); + col2 =its_create_collection (2 /* col id */ ,2 /* target PE */ ); + +gicv3_lpi_set_config (8195 ,LPI_PROP_DEFAULT ); +gicv3_lpi_set_config (8196 ,LPI_PROP_DEFAULT ); + +its_send_invall (col2); +its_send_invall (col3); + +report_prefix_push ("int" ); +/* +* dev=2, eventid=20 -> lpi= 8195, col=3 +* dev=7, eventid=255 -> lpi= 8196, col=2 +* Trigger dev2, eventid=20 and dev7, eventid=255 +* Check both LPIs hit +*/ +
arm/gic.c static void test_its_introspection(void) · line 705
Quoted 10 lines
arm/gic.c +10
-0+its_send_mapd (dev2, true); +its_send_mapd (dev7, true); + +its_send_mapc (col3, true); +its_send_mapc (col2, true); + +its_send_mapti (dev2,8195 /* lpi id */ , +20 /* event id */ , col3); +its_send_mapti (dev7,8196 /* lpi id */ , +255 /* event id */ , col2);
No need for line breaks, with the embedded comments it's hard to read
arm/gic.c static void test_its_introspection(void) · line 715
Quoted 10 lines
arm/gic.c +10
-0+ +lpi_stats_expect (3 ,8195 ); +its_send_int (dev2,20 ); +check_lpi_stats (); + +lpi_stats_expect (2 ,8196 ); +its_send_int (dev7,255 ); +check_lpi_stats (); + +report_prefix_pop ();
I think a table of parameters and loop would be nicer than all the repeated function calls.
arm/gic.c static void test_its_introspection(void) · line 723
Quoted 8 lines
arm/gic.c +8
-0+ +report_prefix_push ("inv/invall" ); + +/* +* disable 8195, check dev2/eventid=20 does not trigger the +* corresponding LPI +*/ +gicv3_lpi_set_config (8195 ,LPI_PROP_DEFAULT & ~0x1 );
LPI_PROP_DEFAULT & ~LPI_PROP_ENABLED
Quoted 13 lines
arm/gic.c +13
-0+its_send_inv (dev2,20 ); + +lpi_stats_expect (-1 ,-1 ); +its_send_int (dev2,20 ); +check_lpi_stats (); + +/* +* re-enable the LPI but willingly do not call invall +* so the change in config is not taken into account. +* The LPI should not hit +*/ +gicv3_lpi_set_config (8195 ,LPI_PROP_DEFAULT ); +lpi_stats_expect (-1 ,-1 );
arm/gic.c static void test_its_introspection(void) · line 746
Quoted 10 lines
arm/gic.c +10
-0+its_send_int (dev2,20 ); +check_lpi_stats (); + +/* Now call the invall and check the LPI hits */ +its_send_invall (col3); +lpi_stats_expect (3 ,8195 ); +its_send_int (dev2,20 ); +check_lpi_stats (); + +report_prefix_pop ();
Need blank line here.
arm/gic.c static void test_its_introspection(void) · line 751
Quoted 5 lines
arm/gic.c +5
-0+/* +* Unmap device 2 and check the eventid 20 formerly +* attached to it does not hit anymore +*/ +report_prefix_push ("mapd valid=false" );
Above you have the prefix-push before the comment explaining the test. After is probably better, but whatever, as long as it's consistent.
Quoted 49 lines
arm/gic.c +18
-0+its_send_mapd (dev2, false); +lpi_stats_expect (-1 ,-1 ); +its_send_int (dev2,20 ); +check_lpi_stats (); +report_prefix_pop (); + +/* Unmap the collection this time and check no LPI does hit */ +report_prefix_push ("mapc valid=false" ); +its_send_mapc (col2, false); +lpi_stats_expect (-1 ,-1 ); +its_send_int (dev7,255 ); +check_lpi_stats (); + report_prefix_pop(); +}#endif int main (int argc,char **argv)@@ -594,6 +796,10 @@ int main(int argc, char **argv)report_prefix_push (argv[1 ]);gic_test_mmio ();report_prefix_pop (); + }else if (!strcmp (argv[1 ],"its-trigger" )) { +report_prefix_push (argv[1 ]); +test_its_trigger (); +report_prefix_pop (); }else if (strcmp (argv[1 ],"its-introspection" ) ==0 ) {report_prefix_push (argv[1 ]);test_its_introspection ();arm/unittests.cfg +7
-0@@ -129,6 +129,13 @@ extra_params = -machine gic-version=3 -append 'its-introspection' groups = its arch = arm64 +[its-trigger] +file = gic.flat +smp = $MAX_SMP +extra_params = -machine gic-version=3 -append 'its-trigger' +groups = its +arch = arm64 + # Test PSCI emulation [psci] file = psci.flatSignature · 1 line
2.20.1
Thanks,
drew
Re: [kvm-unit-tests PATCH v3 11/14] arm/arm64: ITS: INT functional tests
Hi Drew,
Quoted 58 lines · Andrew Jones
Quoted 57 lines · Eric Auger
Triggers LPIs through the INT command.
the test checks the LPI hits the right CPU and triggers the right LPI intid, ie. the translation is correct.
Updates to the config table also are tested, along with inv and invall commands.
Signed-off-by Eric Auger <[email protected]>---
v2 -> v3:
- add comments
- keep thereport_skipin case there aren't 4 vcpus to be able to run other tests in the its category.
- fix the prefix pop
- moveits_eventandits_statstoarm/gic.c--- arm/gic.c | 228 +++++++++++++++++++++++++++++++++++++++++++--- arm/unittests.cfg | 7 ++ 2 files changed, 224 insertions(+), 11 deletions(-)arm/gic.c +26
-0@@ -160,6 +160,87 @@ static void ipi_handler(struct pt_regs *regs __unused) } } +static void setup_irq (handler_t handler) +{ +gic_enable_defaults (); +#ifdef __arm__ +install_exception_handler (EXCPTN_IRQ , handler); +#else +install_irq_handler (EL1H_IRQ , handler); +#endif +local_irq_enable (); +} + +#if defined(__aarch64__) +struct its_event { +int cpu_id; +int lpi_id; +}; + +struct its_stats { +struct its_event expected; +struct its_event observed; +}; + +static struct its_stats lpi_stats; + +static void lpi_handler(struct pt_regs *regs __unused) +{
Quoted 8 lines
arm/gic.c static void ipi_handler(struct pt_regs *regs __unused) · line 194
Quoted 6 lines
arm/gic.c +6
-0+u32 irqstat =gic_read_iar (); +int irqnr =gic_iar_irqnr (irqstat); + +gic_write_eoir (irqstat); +if (irqnr <8192 ) +report (false,"Unexpected non LPI interrupt received" );
report_info
why? This is an error case. We do not expect other interrupts than LPIs
Quoted 30 lines
Quoted 7 lines
arm/gic.c +6
-0+smp_rmb ();/* pairs with wmb in lpi_stats_expect */ + lpi_stats.observed.cpu_id =smp_processor_id (); + lpi_stats.observed.lpi_id = irqnr; +smp_wmb ();/* pairs with rmb in check_lpi_stats */ +} +arm/gic.c static void ipi_handler(struct pt_regs *regs __unused) · line 210
Quoted 10 lines
arm/gic.c +10
-0+static void lpi_stats_expect (int exp_cpu_id,int exp_lpi_id) +{ + lpi_stats.expected.cpu_id = exp_cpu_id; + lpi_stats.expected.lpi_id = exp_lpi_id; + lpi_stats.observed.cpu_id =-1 ; + lpi_stats.observed.lpi_id =-1 ; +smp_wmb ();/* pairs with rmb in handler */ +} + +static void check_lpi_stats (void )static void
check_lpi_stats(const char *testname){ bool pass = false;arm/gic.c static void ipi_handler(struct pt_regs *regs __unused) · line 215
Quoted 5 lines
arm/gic.c +5
-0+{ +mdelay (100 ); +smp_rmb ();/* pairs with wmb in lpi_handler */ +if ((lpi_stats.observed.cpu_id != lpi_stats.expected.cpu_id) || + (lpi_stats.observed.lpi_id != lpi_stats.expected.lpi_id)) {nit: extra ()
Quoted 8 lines
arm/gic.c static void ipi_handler(struct pt_regs *regs __unused) · line 221
Quoted 6 lines
arm/gic.c +6
-0+if (lpi_stats.observed.cpu_id ==-1 && + lpi_stats.observed.lpi_id ==-1 ) { +report (false, +"No LPI received whereas (cpuid=%d, intid=%d) " +"was expected" , lpi_stats.expected.cpu_id, + lpi_stats.expected.lpi_id);
report_info
What's the problem keeping those. Those are error reports. The message is something like that:
FAIL: gicv3: its-trigger: mapc valid=false: No LPI received whereas (cpuid=1, intid=8192) was expected.
So the testname is already part of the message.
Quoted 72 lines
arm/gic.c static void ipi_handler(struct pt_regs *regs __unused) · line 225
Quoted 5 lines
arm/gic.c +4
-0+ }else { +report (false,"Unexpected LPI (cpuid=%d, intid=%d)" , + lpi_stats.observed.cpu_id, + lpi_stats.observed.lpi_id);
report_infoarm/gic.c static void ipi_handler(struct pt_regs *regs __unused) · line 226
Quoted 1 line
arm/gic.c +1
-0+ }pass = false;
arm/gic.c static void ipi_handler(struct pt_regs *regs __unused) · line 229
Quoted 3 lines
arm/gic.c +3
-0+ }else if (lpi_stats.expected.lpi_id !=-1 ) { +report (true,"LPI %d on CPU %d" , lpi_stats.observed.lpi_id, + lpi_stats.observed.cpu_id);
report_infoarm/gic.c static void ipi_handler(struct pt_regs *regs __unused) · line 231
Quoted 2 lines
arm/gic.c +2
-0+ }else { +report (true,"no LPI received, as expected" );
report_infoarm/gic.c static void ipi_handler(struct pt_regs *regs __unused) · line 232
Quoted 1 line
arm/gic.c +1
-0+ }
report(pass, "%s", testname);Quoted 44 lines
arm/gic.c +12
-11+} + +static void secondary_lpi_test (void ) +{ +setup_irq (lpi_handler); +cpumask_set_cpu (smp_processor_id (), &ready); +while (1 ) +wfi (); +} +#endif +static void gicv2_ipi_send_self (void ) {writel (2 <<24 |IPI_IRQ ,gicv2_dist_base () +GICD_SGIR );@@ -217,17 +298,6 @@ static void ipi_test_smp(void)report_prefix_pop (); } -static void setup_irq (handler_t handler) -{ -gic_enable_defaults (); -#ifdef __arm__ -install_exception_handler (EXCPTN_IRQ , handler); -#else -install_irq_handler (EL1H_IRQ , handler); -#endif -local_irq_enable (); -} -static void ipi_send (void ) {setup_irq (ipi_handler);@@ -522,6 +592,7 @@ static void gic_test_mmio(void)#if defined(__arm__)static void test_its_introspection (void ) {} +static void test_its_trigger (void ) {}#else /* __arch64__ */ @@ -561,6 +632,137 @@ static void test_its_introspection(void)report_info ("collection baser entry_size = 0x%x" , coll_baser->esz); }
Quoted 13 lines
arm/gic.c static void test_its_introspection(void) · line 644
Quoted 10 lines
arm/gic.c +10
-0+static bool its_prerequisites (int nb_cpus) +{ +int cpu; + +if (!gicv3_its_base ()) { +report_skip ("No ITS, skip ..." ); +return true; + } + +if (nr_cpus <4 ) {
nr_cpus<nb_cpus, or just drop thenb_cpusparameter and hard code 4 here.
sure
Quoted 16 lines
Quoted 16 lines
arm/gic.c +15
-0+report_skip ("Test requires at least %d vcpus" , nb_cpus); +return true; + } + +stats_reset (); + +setup_irq (lpi_handler); + +for_each_present_cpu (cpu) { +if (cpu ==0 ) +continue ; +smp_boot_secondary (cpu, secondary_lpi_test); + } +wait_on_ready (); +
Quoted 8 lines
arm/gic.c static void test_its_introspection(void) · line 665
Quoted 6 lines
arm/gic.c +6
-0+its_enable_defaults (); + +lpi_stats_expect (-1 ,-1 ); +check_lpi_stats (); + +return false;Reverse logic. I'd expect 'return true' for success.
I am going to return an int. In case of error a std negative error will be returned.
Quoted 38 lines
arm/gic.c static void test_its_introspection(void) · line 673
Quoted 9 lines
arm/gic.c +8
-0+} + +static void test_its_trigger (void ) +{ +struct its_collection *col3, *col2; +struct its_device *dev2, *dev7; + +if (its_prerequisites (4 ))if (!its_prerequisites(...))
Quoted 26 lines
arm/gic.c +26
-0+return ; + + dev2 =its_create_device (2 /* dev id */ ,8 /* nb_ites */ ); + dev7 =its_create_device (7 /* dev id */ ,8 /* nb_ites */ ); + + col3 =its_create_collection (3 /* col id */ ,3 /* target PE */ ); + col2 =its_create_collection (2 /* col id */ ,2 /* target PE */ ); + +gicv3_lpi_set_config (8195 ,LPI_PROP_DEFAULT ); +gicv3_lpi_set_config (8196 ,LPI_PROP_DEFAULT ); + +its_send_invall (col2); +its_send_invall (col3); + +report_prefix_push ("int" ); +/* +* dev=2, eventid=20 -> lpi= 8195, col=3 +* dev=7, eventid=255 -> lpi= 8196, col=2 +* Trigger dev2, eventid=20 and dev7, eventid=255 +* Check both LPIs hit +*/ + +its_send_mapd (dev2, true); +its_send_mapd (dev7, true); + +its_send_mapc (col3, true);
Quoted 8 lines
arm/gic.c static void test_its_introspection(void) · line 705
Quoted 6 lines
arm/gic.c +6
-0+its_send_mapc (col2, true); + +its_send_mapti (dev2,8195 /* lpi id */ , +20 /* event id */ , col3); +its_send_mapti (dev7,8196 /* lpi id */ , +255 /* event id */ , col2);No need for line breaks, with the embedded comments it's hard to read
OK
Quoted 14 lines
arm/gic.c static void test_its_introspection(void) · line 715
Quoted 11 lines
arm/gic.c +10
-0+ +lpi_stats_expect (3 ,8195 ); +its_send_int (dev2,20 ); +check_lpi_stats (); + +lpi_stats_expect (2 ,8196 ); +its_send_int (dev7,255 ); +check_lpi_stats (); + +report_prefix_pop ();I think a table of parameters and loop would be nicer than all the repeated function calls.
Frankly speaking I am not sure this would really help. We are just enabling 2 translation paths. I think I prefer to manipulate the low level objects and helpers rather than playing with a loop and potential new structs of params.
Quoted 11 lines
arm/gic.c static void test_its_introspection(void) · line 723
Quoted 9 lines
arm/gic.c +8
-0+ +report_prefix_push ("inv/invall" ); + +/* +* disable 8195, check dev2/eventid=20 does not trigger the +* corresponding LPI +*/ +gicv3_lpi_set_config (8195 ,LPI_PROP_DEFAULT & ~0x1 );
LPI_PROP_DEFAULT& ~LPI_PROP_ENABLED
ok
Quoted 18 lines
Quoted 18 lines
arm/gic.c +17
-0+its_send_inv (dev2,20 ); + +lpi_stats_expect (-1 ,-1 ); +its_send_int (dev2,20 ); +check_lpi_stats (); + +/* +* re-enable the LPI but willingly do not call invall +* so the change in config is not taken into account. +* The LPI should not hit +*/ +gicv3_lpi_set_config (8195 ,LPI_PROP_DEFAULT ); +lpi_stats_expect (-1 ,-1 ); +its_send_int (dev2,20 ); +check_lpi_stats (); + +/* Now call the invall and check the LPI hits */
Quoted 8 lines
arm/gic.c static void test_its_introspection(void) · line 715
Quoted 6 lines
arm/gic.c +6
-0+its_send_invall (col3); +lpi_stats_expect (3 ,8195 ); +its_send_int (dev2,20 ); +check_lpi_stats (); + +report_prefix_pop ();Need blank line here.
OK
Quoted 9 lines
arm/gic.c static void test_its_introspection(void) · line 751
Quoted 6 lines
arm/gic.c +5
-0+/* +* Unmap device 2 and check the eventid 20 formerly +* attached to it does not hit anymore +*/ +report_prefix_push ("mapd valid=false" );Above you have the prefix-push before the comment explaining the test. After is probably better, but whatever, as long as it's consistent.
moved after
Quoted 37 lines
Quoted 37 lines
arm/gic.c +18
-0+its_send_mapd (dev2, false); +lpi_stats_expect (-1 ,-1 ); +its_send_int (dev2,20 ); +check_lpi_stats (); +report_prefix_pop (); + +/* Unmap the collection this time and check no LPI does hit */ +report_prefix_push ("mapc valid=false" ); +its_send_mapc (col2, false); +lpi_stats_expect (-1 ,-1 ); +its_send_int (dev7,255 ); +check_lpi_stats (); + report_prefix_pop(); +}#endif int main (int argc,char **argv)@@ -594,6 +796,10 @@ int main(int argc, char **argv)report_prefix_push (argv[1 ]);gic_test_mmio ();report_prefix_pop (); + }else if (!strcmp (argv[1 ],"its-trigger" )) { +report_prefix_push (argv[1 ]); +test_its_trigger (); +report_prefix_pop (); }else if (strcmp (argv[1 ],"its-introspection" ) ==0 ) {report_prefix_push (argv[1 ]);test_its_introspection ();arm/unittests.cfg +0
-0@@ -129,6 +129,13 @@ extra_params = -machine gic-version=3 -append 'its-introspection' groups = its arch = arm64
Quoted 17 lines
Quoted 13 lines
arm/unittests.cfg +7
-0+[its-trigger] +file = gic.flat +smp = $MAX_SMP +extra_params = -machine gic-version=3 -append 'its-trigger' +groups = its +arch = arm64 + # Test PSCI emulation [psci] file = psci.flatSignature · 1 line
2.20.1Thanks,
drew
Thanks
Eric
Re: [kvm-unit-tests PATCH v3 11/14] arm/arm64: ITS: INT functional tests
Quoted 69 lines · Auger Eric
Hi Drew,
Quoted 58 lines · Andrew Jones
Quoted 57 lines · Eric Auger
Triggers LPIs through the INT command.
the test checks the LPI hits the right CPU and triggers the right LPI intid, ie. the translation is correct.
Updates to the config table also are tested, along with inv and invall commands.
Signed-off-by Eric Auger <[email protected]>---
v2 -> v3:
- add comments
- keep thereport_skipin case there aren't 4 vcpus to be able to run other tests in the its category.
- fix the prefix pop
- moveits_eventandits_statstoarm/gic.c--- arm/gic.c | 228 +++++++++++++++++++++++++++++++++++++++++++--- arm/unittests.cfg | 7 ++ 2 files changed, 224 insertions(+), 11 deletions(-)arm/gic.c +26
-0@@ -160,6 +160,87 @@ static void ipi_handler(struct pt_regs *regs __unused) } } +static void setup_irq (handler_t handler) +{ +gic_enable_defaults (); +#ifdef __arm__ +install_exception_handler (EXCPTN_IRQ , handler); +#else +install_irq_handler (EL1H_IRQ , handler); +#endif +local_irq_enable (); +} + +#if defined(__aarch64__) +struct its_event { +int cpu_id; +int lpi_id; +}; + +struct its_stats { +struct its_event expected; +struct its_event observed; +}; + +static struct its_stats lpi_stats; + +static void lpi_handler(struct pt_regs *regs __unused) +{Quoted 8 lines
arm/gic.c static void ipi_handler(struct pt_regs *regs __unused) · line 194
Quoted 6 lines
arm/gic.c +6
-0+u32 irqstat =gic_read_iar (); +int irqnr =gic_iar_irqnr (irqstat); + +gic_write_eoir (irqstat); +if (irqnr <8192 ) +report (false,"Unexpected non LPI interrupt received" );
report_info
Quoted 1 line
why? This is an error case. We do not expect other interrupts than LPIs
If there's almost no chance this will happen and it means something quite unexpected has occurred, then it should probably be an assert. If this is a real test case, then it should be
report(irqnr >= 8192, "Got LPI");
or something like that. If it's something that shouldn't ever happen, so it doesn't really deserve its own PASS/FAIL test output each execution of the unit test, but you don't want to assert for some reason, then it should be a report_info, but it should probably also contain a "WARNING" prefix in that case.
Quoted 38 lines
Quoted 30 lines
Quoted 7 lines
arm/gic.c +6
-0+smp_rmb ();/* pairs with wmb in lpi_stats_expect */ + lpi_stats.observed.cpu_id =smp_processor_id (); + lpi_stats.observed.lpi_id = irqnr; +smp_wmb ();/* pairs with rmb in check_lpi_stats */ +} +arm/gic.c static void ipi_handler(struct pt_regs *regs __unused) · line 210
Quoted 10 lines
arm/gic.c +10
-0+static void lpi_stats_expect (int exp_cpu_id,int exp_lpi_id) +{ + lpi_stats.expected.cpu_id = exp_cpu_id; + lpi_stats.expected.lpi_id = exp_lpi_id; + lpi_stats.observed.cpu_id =-1 ; + lpi_stats.observed.lpi_id =-1 ; +smp_wmb ();/* pairs with rmb in handler */ +} + +static void check_lpi_stats (void )static void
check_lpi_stats(const char *testname){ bool pass = false;arm/gic.c static void ipi_handler(struct pt_regs *regs __unused) · line 215
Quoted 5 lines
arm/gic.c +5
-0+{ +mdelay (100 ); +smp_rmb ();/* pairs with wmb in lpi_handler */ +if ((lpi_stats.observed.cpu_id != lpi_stats.expected.cpu_id) || + (lpi_stats.observed.lpi_id != lpi_stats.expected.lpi_id)) {nit: extra ()
Quoted 8 lines
arm/gic.c static void ipi_handler(struct pt_regs *regs __unused) · line 221
Quoted 6 lines
arm/gic.c +6
-0+if (lpi_stats.observed.cpu_id ==-1 && + lpi_stats.observed.lpi_id ==-1 ) { +report (false, +"No LPI received whereas (cpuid=%d, intid=%d) " +"was expected" , lpi_stats.expected.cpu_id, + lpi_stats.expected.lpi_id);
report_info
Quoted 6 lines
What's the problem keeping those. Those are error reports. The message is something like that:
FAIL: gicv3: its-trigger: mapc valid=false: No LPI received whereas (cpuid=1, intid=8192) was expected.So the testname is already part of the message.
This one has two problems with being report() vs. report_info. The same comment as above, where the condition for a report() should be the test, rather than if (cond) report(false, ...), which implies it's not expected to report at all. A pattern like that needs to be extended at least to something like this
if (cond)
report(true, ...)
else
report(false, ...)
so we get the PASS/FAIL each execution. The other problem with this particular report() is the dynamic info in it (cpuid and maybe intid). A report() should only have consistent info so test output parsers can count on finding the PASS/FAIL for a given report line. If you need a test like this, then it can be structured like
report_info(...); // dynamic info
if (cond) {
report(true, MSG1); // no dynamic info
report(true, MSG2); // no dynamic info
} else {
report(false, MSG1); // no dynamic info
report(false, MSG2); // no dynamic info
}
Notice how the MSG's match on both paths of the condition.
Or just
report_info(...);report(cond, ...);
Quoted 18 lines
Quoted 18 lines
arm/gic.c static void ipi_handler(struct pt_regs *regs __unused) · line 225
Quoted 5 lines
arm/gic.c +4
-0+ }else { +report (false,"Unexpected LPI (cpuid=%d, intid=%d)" , + lpi_stats.observed.cpu_id, + lpi_stats.observed.lpi_id);
report_infoarm/gic.c static void ipi_handler(struct pt_regs *regs __unused) · line 226
Quoted 1 line
arm/gic.c +1
-0+ }pass = false;
arm/gic.c static void ipi_handler(struct pt_regs *regs __unused) · line 229
Quoted 3 lines
arm/gic.c +3
-0+ }else if (lpi_stats.expected.lpi_id !=-1 ) { +report (true,"LPI %d on CPU %d" , lpi_stats.observed.lpi_id, + lpi_stats.observed.cpu_id);
report_info
Quoted 4 lines
Quoted 4 lines
arm/gic.c static void ipi_handler(struct pt_regs *regs __unused) · line 231
Quoted 2 lines
arm/gic.c +2
-0+ }else { +report (true,"no LPI received, as expected" );
report_info
This if, else if, ..., else with report() would be fine if the messages would all match, resulting in a single 'PASS: MSG' line. report_info can be used to get the dynamic info output too.
Quoted 151 lines
Quoted 50 lines
arm/gic.c static void ipi_handler(struct pt_regs *regs __unused) · line 232
Quoted 3 lines
arm/gic.c +1
-0+ }
report(pass, "%s", testname);Quoted 44 lines
arm/gic.c +12
-11+} + +static void secondary_lpi_test (void ) +{ +setup_irq (lpi_handler); +cpumask_set_cpu (smp_processor_id (), &ready); +while (1 ) +wfi (); +} +#endif +static void gicv2_ipi_send_self (void ) {writel (2 <<24 |IPI_IRQ ,gicv2_dist_base () +GICD_SGIR );@@ -217,17 +298,6 @@ static void ipi_test_smp(void)report_prefix_pop (); } -static void setup_irq (handler_t handler) -{ -gic_enable_defaults (); -#ifdef __arm__ -install_exception_handler (EXCPTN_IRQ , handler); -#else -install_irq_handler (EL1H_IRQ , handler); -#endif -local_irq_enable (); -} -static void ipi_send (void ) {setup_irq (ipi_handler);@@ -522,6 +592,7 @@ static void gic_test_mmio(void)#if defined(__arm__)static void test_its_introspection (void ) {} +static void test_its_trigger (void ) {}#else /* __arch64__ */ @@ -561,6 +632,137 @@ static void test_its_introspection(void)report_info ("collection baser entry_size = 0x%x" , coll_baser->esz); }Quoted 13 lines
arm/gic.c static void test_its_introspection(void) · line 644
Quoted 10 lines
arm/gic.c +10
-0+static bool its_prerequisites (int nb_cpus) +{ +int cpu; + +if (!gicv3_its_base ()) { +report_skip ("No ITS, skip ..." ); +return true; + } + +if (nr_cpus <4 ) {
nr_cpus<nb_cpus, or just drop thenb_cpusparameter and hard code 4 here.sure
Quoted 16 lines
Quoted 16 lines
arm/gic.c +15
-0+report_skip ("Test requires at least %d vcpus" , nb_cpus); +return true; + } + +stats_reset (); + +setup_irq (lpi_handler); + +for_each_present_cpu (cpu) { +if (cpu ==0 ) +continue ; +smp_boot_secondary (cpu, secondary_lpi_test); + } +wait_on_ready (); +Quoted 8 lines
arm/gic.c static void test_its_introspection(void) · line 665
Quoted 6 lines
arm/gic.c +6
-0+its_enable_defaults (); + +lpi_stats_expect (-1 ,-1 ); +check_lpi_stats (); + +return false;Reverse logic. I'd expect 'return true' for success.
I am going to return an int. In case of error a std negative error will be returned.
Quoted 38 lines
arm/gic.c static void test_its_introspection(void) · line 673
Quoted 9 lines
arm/gic.c +8
-0+} + +static void test_its_trigger (void ) +{ +struct its_collection *col3, *col2; +struct its_device *dev2, *dev7; + +if (its_prerequisites (4 ))if (!its_prerequisites(...))
Quoted 26 lines
arm/gic.c +26
-0+return ; + + dev2 =its_create_device (2 /* dev id */ ,8 /* nb_ites */ ); + dev7 =its_create_device (7 /* dev id */ ,8 /* nb_ites */ ); + + col3 =its_create_collection (3 /* col id */ ,3 /* target PE */ ); + col2 =its_create_collection (2 /* col id */ ,2 /* target PE */ ); + +gicv3_lpi_set_config (8195 ,LPI_PROP_DEFAULT ); +gicv3_lpi_set_config (8196 ,LPI_PROP_DEFAULT ); + +its_send_invall (col2); +its_send_invall (col3); + +report_prefix_push ("int" ); +/* +* dev=2, eventid=20 -> lpi= 8195, col=3 +* dev=7, eventid=255 -> lpi= 8196, col=2 +* Trigger dev2, eventid=20 and dev7, eventid=255 +* Check both LPIs hit +*/ + +its_send_mapd (dev2, true); +its_send_mapd (dev7, true); + +its_send_mapc (col3, true);Quoted 8 lines
arm/gic.c static void test_its_introspection(void) · line 705
Quoted 6 lines
arm/gic.c +6
-0+its_send_mapc (col2, true); + +its_send_mapti (dev2,8195 /* lpi id */ , +20 /* event id */ , col3); +its_send_mapti (dev7,8196 /* lpi id */ , +255 /* event id */ , col2);No need for line breaks, with the embedded comments it's hard to read
OK
Quoted 14 lines
arm/gic.c static void test_its_introspection(void) · line 715
Quoted 11 lines
arm/gic.c +10
-0+ +lpi_stats_expect (3 ,8195 ); +its_send_int (dev2,20 ); +check_lpi_stats (); + +lpi_stats_expect (2 ,8196 ); +its_send_int (dev7,255 ); +check_lpi_stats (); + +report_prefix_pop ();I think a table of parameters and loop would be nicer than all the repeated function calls.
Quoted 4 lines
Frankly speaking I am not sure this would really help. We are just enabling 2 translation paths. I think I prefer to manipulate the low level objects and helpers rather than playing with a loop and potential new structs of params.
OK, but you could probably at least wrap the common sequence into one function
void master_function(a1, a2, a3, a4)
{
lpi_stats_expect(a1, a2);
its_send_int(a3, a4);
check_lpi_stats();
}
but whatever, it's not so important.
Quoted 108 lines
Quoted 11 lines
arm/gic.c static void test_its_introspection(void) · line 723
Quoted 9 lines
arm/gic.c +8
-0+ +report_prefix_push ("inv/invall" ); + +/* +* disable 8195, check dev2/eventid=20 does not trigger the +* corresponding LPI +*/ +gicv3_lpi_set_config (8195 ,LPI_PROP_DEFAULT & ~0x1 );
LPI_PROP_DEFAULT& ~LPI_PROP_ENABLEDok
Quoted 18 lines
Quoted 18 lines
arm/gic.c +17
-0+its_send_inv (dev2,20 ); + +lpi_stats_expect (-1 ,-1 ); +its_send_int (dev2,20 ); +check_lpi_stats (); + +/* +* re-enable the LPI but willingly do not call invall +* so the change in config is not taken into account. +* The LPI should not hit +*/ +gicv3_lpi_set_config (8195 ,LPI_PROP_DEFAULT ); +lpi_stats_expect (-1 ,-1 ); +its_send_int (dev2,20 ); +check_lpi_stats (); + +/* Now call the invall and check the LPI hits */ Quoted 8 lines
arm/gic.c static void test_its_introspection(void) · line 715
Quoted 6 lines
arm/gic.c +6
-0+its_send_invall (col3); +lpi_stats_expect (3 ,8195 ); +its_send_int (dev2,20 ); +check_lpi_stats (); + +report_prefix_pop ();Need blank line here.
OK
Quoted 9 lines
arm/gic.c static void test_its_introspection(void) · line 751
Quoted 6 lines
arm/gic.c +5
-0+/* +* Unmap device 2 and check the eventid 20 formerly +* attached to it does not hit anymore +*/ +report_prefix_push ("mapd valid=false" );Above you have the prefix-push before the comment explaining the test. After is probably better, but whatever, as long as it's consistent.
moved after
Quoted 37 lines
Quoted 37 lines
arm/gic.c +18
-0+its_send_mapd (dev2, false); +lpi_stats_expect (-1 ,-1 ); +its_send_int (dev2,20 ); +check_lpi_stats (); +report_prefix_pop (); + +/* Unmap the collection this time and check no LPI does hit */ +report_prefix_push ("mapc valid=false" ); +its_send_mapc (col2, false); +lpi_stats_expect (-1 ,-1 ); +its_send_int (dev7,255 ); +check_lpi_stats (); + report_prefix_pop(); +}#endif int main (int argc,char **argv)@@ -594,6 +796,10 @@ int main(int argc, char **argv)report_prefix_push (argv[1 ]);gic_test_mmio ();report_prefix_pop (); + }else if (!strcmp (argv[1 ],"its-trigger" )) { +report_prefix_push (argv[1 ]); +test_its_trigger (); +report_prefix_pop (); }else if (strcmp (argv[1 ],"its-introspection" ) ==0 ) {report_prefix_push (argv[1 ]);test_its_introspection ();arm/unittests.cfg +0
-0@@ -129,6 +129,13 @@ extra_params = -machine gic-version=3 -append 'its-introspection' groups = its arch = arm64Quoted 17 lines
Quoted 13 lines
arm/unittests.cfg +7
-0+[its-trigger] +file = gic.flat +smp = $MAX_SMP +extra_params = -machine gic-version=3 -append 'its-trigger' +groups = its +arch = arm64 + # Test PSCI emulation [psci] file = psci.flatSignature · 1 line
2.20.1Thanks,
drewThanks
Eric
Re: [kvm-unit-tests PATCH v3 11/14] arm/arm64: ITS: INT functional tests
Hi Drew,
Quoted 76 lines · Andrew Jones
Quoted 69 lines · Auger Eric
Hi Drew,
Quoted 58 lines · Andrew Jones
Quoted 57 lines · Eric Auger
Triggers LPIs through the INT command.
the test checks the LPI hits the right CPU and triggers the right LPI intid, ie. the translation is correct.
Updates to the config table also are tested, along with inv and invall commands.
Signed-off-by Eric Auger <[email protected]>---
v2 -> v3:
- add comments
- keep thereport_skipin case there aren't 4 vcpus to be able to run other tests in the its category.
- fix the prefix pop
- moveits_eventandits_statstoarm/gic.c--- arm/gic.c | 228 +++++++++++++++++++++++++++++++++++++++++++--- arm/unittests.cfg | 7 ++ 2 files changed, 224 insertions(+), 11 deletions(-)arm/gic.c +26
-0@@ -160,6 +160,87 @@ static void ipi_handler(struct pt_regs *regs __unused) } } +static void setup_irq (handler_t handler) +{ +gic_enable_defaults (); +#ifdef __arm__ +install_exception_handler (EXCPTN_IRQ , handler); +#else +install_irq_handler (EL1H_IRQ , handler); +#endif +local_irq_enable (); +} + +#if defined(__aarch64__) +struct its_event { +int cpu_id; +int lpi_id; +}; + +struct its_stats { +struct its_event expected; +struct its_event observed; +}; + +static struct its_stats lpi_stats; + +static void lpi_handler(struct pt_regs *regs __unused) +{Quoted 8 lines
arm/gic.c static void ipi_handler(struct pt_regs *regs __unused) · line 194
Quoted 6 lines
arm/gic.c +6
-0+u32 irqstat =gic_read_iar (); +int irqnr =gic_iar_irqnr (irqstat); + +gic_write_eoir (irqstat); +if (irqnr <8192 ) +report (false,"Unexpected non LPI interrupt received" );
report_infoQuoted 1 line
why? This is an error case. We do not expect other interrupts than LPIs
If there's almost no chance this will happen and it means something quite unexpected has occurred, then it should probably be an assert. If this is a real test case, then it should be
Quoted 7 lines
report(irqnr >= 8192, "Got LPI");or something like that. If it's something that shouldn't ever happen, so it doesn't really deserve its own PASS/FAIL test output each execution of the unit test, but you don't want to assert for some reason, then it should be a
report_info, but it should probably also contain a "WARNING" prefix in that case.
OK so the assert should be OK.
Quoted 52 lines
Quoted 39 lines
Quoted 31 lines
Quoted 8 lines
arm/gic.c +6
-0+smp_rmb ();/* pairs with wmb in lpi_stats_expect */ + lpi_stats.observed.cpu_id =smp_processor_id (); + lpi_stats.observed.lpi_id = irqnr; +smp_wmb ();/* pairs with rmb in check_lpi_stats */ +} +arm/gic.c static void ipi_handler(struct pt_regs *regs __unused) · line 210
Quoted 10 lines
arm/gic.c +10
-0+static void lpi_stats_expect (int exp_cpu_id,int exp_lpi_id) +{ + lpi_stats.expected.cpu_id = exp_cpu_id; + lpi_stats.expected.lpi_id = exp_lpi_id; + lpi_stats.observed.cpu_id =-1 ; + lpi_stats.observed.lpi_id =-1 ; +smp_wmb ();/* pairs with rmb in handler */ +} + +static void check_lpi_stats (void )static void
check_lpi_stats(const char *testname){ bool pass = false;arm/gic.c static void ipi_handler(struct pt_regs *regs __unused) · line 215
Quoted 5 lines
arm/gic.c +5
-0+{ +mdelay (100 ); +smp_rmb ();/* pairs with wmb in lpi_handler */ +if ((lpi_stats.observed.cpu_id != lpi_stats.expected.cpu_id) || + (lpi_stats.observed.lpi_id != lpi_stats.expected.lpi_id)) {nit: extra ()
Quoted 8 lines
arm/gic.c static void ipi_handler(struct pt_regs *regs __unused) · line 221
Quoted 6 lines
arm/gic.c +6
-0+if (lpi_stats.observed.cpu_id ==-1 && + lpi_stats.observed.lpi_id ==-1 ) { +report (false, +"No LPI received whereas (cpuid=%d, intid=%d) " +"was expected" , lpi_stats.expected.cpu_id, + lpi_stats.expected.lpi_id);
report_infoQuoted 6 lines
What's the problem keeping those. Those are error reports. The message is something like that:
FAIL: gicv3: its-trigger: mapc valid=false: No LPI received whereas (cpuid=1, intid=8192) was expected.So the testname is already part of the message.
This one has two problems with being
report()vs.report_info. The same comment as above, where the condition for areport()should be the test, rather than if (cond)report(false, ...), which implies it's not expected to report at all. A pattern like that needs to be extended at least to something like this
Quoted 4 lines
if (cond)
report(true, ...)
else
report(false, ...)
OK understood. I should have use the test as the 1st param.
Quoted 16 lines
so we get the PASS/FAIL each execution. The other problem with this particular
report()is the dynamic info in it (cpuid and maybe intid). Areport()should only have consistent info so test output parsers can count on finding the PASS/FAIL for a given report line. If you need a test like this, then it can be structured like
report_info(...); // dynamic info
if (cond) {
report(true, MSG1); // no dynamic info
report(true, MSG2); // no dynamic info
} else {
report(false, MSG1); // no dynamic info
report(false, MSG2); // no dynamic info
}
Quoted 6 lines
Notice how the MSG's match on both paths of the condition.
Or just
report_info(...);report(cond, ...);
OK I see what you mean now. I will rewrite it accordingly.
Thank you for the extra explanation
Eric
Quoted 305 lines
Quoted 19 lines
Quoted 19 lines
arm/gic.c static void ipi_handler(struct pt_regs *regs __unused) · line 225
Quoted 6 lines
arm/gic.c +4
-0+ }else { +report (false,"Unexpected LPI (cpuid=%d, intid=%d)" , + lpi_stats.observed.cpu_id, + lpi_stats.observed.lpi_id);
report_infoarm/gic.c static void ipi_handler(struct pt_regs *regs __unused) · line 226
Quoted 1 line
arm/gic.c +1
-0+ }pass = false;
arm/gic.c static void ipi_handler(struct pt_regs *regs __unused) · line 229
Quoted 3 lines
arm/gic.c +3
-0+ }else if (lpi_stats.expected.lpi_id !=-1 ) { +report (true,"LPI %d on CPU %d" , lpi_stats.observed.lpi_id, + lpi_stats.observed.cpu_id);
report_infoQuoted 4 lines
Quoted 4 lines
arm/gic.c static void ipi_handler(struct pt_regs *regs __unused) · line 231
Quoted 2 lines
arm/gic.c +2
-0+ }else { +report (true,"no LPI received, as expected" );
report_infoThis if, else if, ..., else with
report()would be fine if the messages would all match, resulting in a single 'PASS: MSG' line.report_infocan be used to get the dynamic info output too.Quoted 151 lines
Quoted 50 lines
arm/gic.c static void ipi_handler(struct pt_regs *regs __unused) · line 232
Quoted 3 lines
arm/gic.c +1
-0+ }
report(pass, "%s", testname);Quoted 44 lines
arm/gic.c +12
-11+} + +static void secondary_lpi_test (void ) +{ +setup_irq (lpi_handler); +cpumask_set_cpu (smp_processor_id (), &ready); +while (1 ) +wfi (); +} +#endif +static void gicv2_ipi_send_self (void ) {writel (2 <<24 |IPI_IRQ ,gicv2_dist_base () +GICD_SGIR );@@ -217,17 +298,6 @@ static void ipi_test_smp(void)report_prefix_pop (); } -static void setup_irq (handler_t handler) -{ -gic_enable_defaults (); -#ifdef __arm__ -install_exception_handler (EXCPTN_IRQ , handler); -#else -install_irq_handler (EL1H_IRQ , handler); -#endif -local_irq_enable (); -} -static void ipi_send (void ) {setup_irq (ipi_handler);@@ -522,6 +592,7 @@ static void gic_test_mmio(void)#if defined(__arm__)static void test_its_introspection (void ) {} +static void test_its_trigger (void ) {}#else /* __arch64__ */ @@ -561,6 +632,137 @@ static void test_its_introspection(void)report_info ("collection baser entry_size = 0x%x" , coll_baser->esz); }Quoted 13 lines
arm/gic.c static void test_its_introspection(void) · line 644
Quoted 10 lines
arm/gic.c +10
-0+static bool its_prerequisites (int nb_cpus) +{ +int cpu; + +if (!gicv3_its_base ()) { +report_skip ("No ITS, skip ..." ); +return true; + } + +if (nr_cpus <4 ) {
nr_cpus<nb_cpus, or just drop thenb_cpusparameter and hard code 4 here.sure
Quoted 16 lines
Quoted 16 lines
arm/gic.c +15
-0+report_skip ("Test requires at least %d vcpus" , nb_cpus); +return true; + } + +stats_reset (); + +setup_irq (lpi_handler); + +for_each_present_cpu (cpu) { +if (cpu ==0 ) +continue ; +smp_boot_secondary (cpu, secondary_lpi_test); + } +wait_on_ready (); +Quoted 8 lines
arm/gic.c static void test_its_introspection(void) · line 665
Quoted 6 lines
arm/gic.c +6
-0+its_enable_defaults (); + +lpi_stats_expect (-1 ,-1 ); +check_lpi_stats (); + +return false;Reverse logic. I'd expect 'return true' for success.
I am going to return an int. In case of error a std negative error will be returned.
Quoted 38 lines
arm/gic.c static void test_its_introspection(void) · line 673
Quoted 9 lines
arm/gic.c +8
-0+} + +static void test_its_trigger (void ) +{ +struct its_collection *col3, *col2; +struct its_device *dev2, *dev7; + +if (its_prerequisites (4 ))if (!its_prerequisites(...))
Quoted 26 lines
arm/gic.c +26
-0+return ; + + dev2 =its_create_device (2 /* dev id */ ,8 /* nb_ites */ ); + dev7 =its_create_device (7 /* dev id */ ,8 /* nb_ites */ ); + + col3 =its_create_collection (3 /* col id */ ,3 /* target PE */ ); + col2 =its_create_collection (2 /* col id */ ,2 /* target PE */ ); + +gicv3_lpi_set_config (8195 ,LPI_PROP_DEFAULT ); +gicv3_lpi_set_config (8196 ,LPI_PROP_DEFAULT ); + +its_send_invall (col2); +its_send_invall (col3); + +report_prefix_push ("int" ); +/* +* dev=2, eventid=20 -> lpi= 8195, col=3 +* dev=7, eventid=255 -> lpi= 8196, col=2 +* Trigger dev2, eventid=20 and dev7, eventid=255 +* Check both LPIs hit +*/ + +its_send_mapd (dev2, true); +its_send_mapd (dev7, true); + +its_send_mapc (col3, true);Quoted 8 lines
arm/gic.c static void test_its_introspection(void) · line 705
Quoted 6 lines
arm/gic.c +6
-0+its_send_mapc (col2, true); + +its_send_mapti (dev2,8195 /* lpi id */ , +20 /* event id */ , col3); +its_send_mapti (dev7,8196 /* lpi id */ , +255 /* event id */ , col2);No need for line breaks, with the embedded comments it's hard to read
OK
Quoted 14 lines
arm/gic.c static void test_its_introspection(void) · line 715
Quoted 11 lines
arm/gic.c +10
-0+ +lpi_stats_expect (3 ,8195 ); +its_send_int (dev2,20 ); +check_lpi_stats (); + +lpi_stats_expect (2 ,8196 ); +its_send_int (dev7,255 ); +check_lpi_stats (); + +report_prefix_pop ();I think a table of parameters and loop would be nicer than all the repeated function calls.
Quoted 4 lines
Frankly speaking I am not sure this would really help. We are just enabling 2 translation paths. I think I prefer to manipulate the low level objects and helpers rather than playing with a loop and potential new structs of params.
OK, but you could probably at least wrap the common sequence into one function
void
master_function(a1, a2, a3, a4)
{
lpi_stats_expect(a1, a2);
its_send_int(a3, a4);
check_lpi_stats();
}but whatever, it's not so important.
Quoted 108 lines
Quoted 11 lines
arm/gic.c static void test_its_introspection(void) · line 723
Quoted 9 lines
arm/gic.c +8
-0+ +report_prefix_push ("inv/invall" ); + +/* +* disable 8195, check dev2/eventid=20 does not trigger the +* corresponding LPI +*/ +gicv3_lpi_set_config (8195 ,LPI_PROP_DEFAULT & ~0x1 );
LPI_PROP_DEFAULT& ~LPI_PROP_ENABLEDok
Quoted 18 lines
Quoted 18 lines
arm/gic.c +17
-0+its_send_inv (dev2,20 ); + +lpi_stats_expect (-1 ,-1 ); +its_send_int (dev2,20 ); +check_lpi_stats (); + +/* +* re-enable the LPI but willingly do not call invall +* so the change in config is not taken into account. +* The LPI should not hit +*/ +gicv3_lpi_set_config (8195 ,LPI_PROP_DEFAULT ); +lpi_stats_expect (-1 ,-1 ); +its_send_int (dev2,20 ); +check_lpi_stats (); + +/* Now call the invall and check the LPI hits */ Quoted 8 lines
arm/gic.c static void test_its_introspection(void) · line 715
Quoted 6 lines
arm/gic.c +6
-0+its_send_invall (col3); +lpi_stats_expect (3 ,8195 ); +its_send_int (dev2,20 ); +check_lpi_stats (); + +report_prefix_pop ();Need blank line here.
OK
Quoted 9 lines
arm/gic.c static void test_its_introspection(void) · line 751
Quoted 6 lines
arm/gic.c +5
-0+/* +* Unmap device 2 and check the eventid 20 formerly +* attached to it does not hit anymore +*/ +report_prefix_push ("mapd valid=false" );Above you have the prefix-push before the comment explaining the test. After is probably better, but whatever, as long as it's consistent.
moved after
Quoted 37 lines
Quoted 37 lines
arm/gic.c +18
-0+its_send_mapd (dev2, false); +lpi_stats_expect (-1 ,-1 ); +its_send_int (dev2,20 ); +check_lpi_stats (); +report_prefix_pop (); + +/* Unmap the collection this time and check no LPI does hit */ +report_prefix_push ("mapc valid=false" ); +its_send_mapc (col2, false); +lpi_stats_expect (-1 ,-1 ); +its_send_int (dev7,255 ); +check_lpi_stats (); + report_prefix_pop(); +}#endif int main (int argc,char **argv)@@ -594,6 +796,10 @@ int main(int argc, char **argv)report_prefix_push (argv[1 ]);gic_test_mmio ();report_prefix_pop (); + }else if (!strcmp (argv[1 ],"its-trigger" )) { +report_prefix_push (argv[1 ]); +test_its_trigger (); +report_prefix_pop (); }else if (strcmp (argv[1 ],"its-introspection" ) ==0 ) {report_prefix_push (argv[1 ]);test_its_introspection ();arm/unittests.cfg +0
-0@@ -129,6 +129,13 @@ extra_params = -machine gic-version=3 -append 'its-introspection' groups = its arch = arm64Quoted 17 lines
Quoted 13 lines
arm/unittests.cfg +7
-0+[its-trigger] +file = gic.flat +smp = $MAX_SMP +extra_params = -machine gic-version=3 -append 'its-trigger' +groups = its +arch = arm64 + # Test PSCI emulation [psci] file = psci.flatSignature · 1 line
2.20.1Thanks,
drewThanks
Eric
PATCH v3 12/14 arm/run: Allow Migration tests
In reply to Eric AugerLet's link getchar.o to use puts and getchar from the tests.
Then allow tests belonging to the migration group to trigger the migration from the test code by putting "migrate" into the uart. Then the code can wait for the migration completion by using getchar().
The __getchar implement is minimalist as it just reads the data register. It is just meant to read the single character emitted at the end of the migration by the runner script.
It is not meant to read more data (FIFOs are not enabled).
---
v2 -> v3:
- take the lock
- assert if more than 16 chars
- removed Thomas' R-b
--- arm/Makefile.common | 2 +- arm/run | 2 +- lib/arm/io.c | 28 ++++++++++++++++++++++++++++ 3 files changed, 30 insertions(+), 2 deletions(-)
arm/Makefile.common +1 -1
@@ -32,7 +32,7 @@ CFLAGS += -I $(SRCDIR)/lib -I $(SRCDIR)/lib/libfdt -I lib asm-offsets = lib/$(ARCH)/asm-offsets.h include $(SRCDIR)/scripts/asm-offsets.mak -cflatobjs += lib/util.o +cflatobjs += lib/util.o lib/getchar.o cflatobjs += lib/alloc_phys.o cflatobjs += lib/alloc_page.o cflatobjs += lib/vmalloc.o
arm/run +1 -1
@@ -61,6 +61,6 @@ fi M+=",accel=$ACCEL" command="$qemu -nodefaults $M -cpu $processor $chr_testdev $pci_testdev" command+=" -display none -serial stdio -kernel" -command="$(timeout_cmd) $command" +command="$(migration_cmd) $(timeout_cmd) $command" run_qemu $command "$@"
lib/arm/io.c +28 -0
@@ -87,6 +87,34 @@ void puts(const char *s)spin_unlock (&uart_lock); } +static int ____getchar (void ) +{ +int c; + +spin_lock (&uart_lock); + c =readb (uart0_base); +spin_unlock (&uart_lock); + +return c ? :-1 ; +} + +/* +* Minimalist implementation for migration completion detection. +* Without FIFOs enabled on the QEMU UART device we just read +* the data register: we cannot read more than 16 characters. +*/ +int __getchar (void ) +{ +int c =____getchar (); +static int count; + +if (c !=-1 ) + ++count; + +assert (count <16 ); + +return c; +}/* * Defining halt to take 'code' as an argument guarantees that it will
Signature · 1 line
2.20.1
Re: [kvm-unit-tests PATCH v3 12/14] arm/run: Allow Migration tests
Quoted 58 lines · Eric Auger
Let's link getchar.o to use puts and getchar from the tests.
Then allow tests belonging to the migration group to trigger the migration from the test code by putting "migrate" into the uart. Then the code can wait for the migration completion by using
getchar().The
__getcharimplement is minimalist as it just reads the data register. It is just meant to read the single character emitted at the end of the migration by the runner script.It is not meant to read more data (FIFOs are not enabled).
Signed-off-by Eric Auger <[email protected]>---
v2 -> v3:
- take the lock
- assert if more than 16 chars
- removed Thomas' R-b--- arm/Makefile.common | 2 +- arm/run | 2 +- lib/arm/io.c | 28 ++++++++++++++++++++++++++++ 3 files changed, 30 insertions(+), 2 deletions(-)arm/Makefile.common +1
-1@@ -32,7 +32,7 @@ CFLAGS += -I $(SRCDIR)/lib -I $(SRCDIR)/lib/libfdt -I lib asm-offsets = lib/$(ARCH)/asm-offsets.h include $(SRCDIR)/scripts/asm-offsets.mak -cflatobjs += lib/util.o +cflatobjs += lib/util.o lib/getchar.o cflatobjs += lib/alloc_phys.o cflatobjs += lib/alloc_page.o cflatobjs += lib/vmalloc.oarm/run +1
-1@@ -61,6 +61,6 @@ fi M+=",accel=$ACCEL" command="$qemu -nodefaults $M -cpu $processor $chr_testdev $pci_testdev" command+=" -display none -serial stdio -kernel" -command="$(timeout_cmd) $command" +command="$(migration_cmd) $(timeout_cmd) $command" run_qemu $command "$@"lib/arm/io.c +0
-0
lib/arm/io.c void puts(const char *s) · line 90
Quoted 4 lines
lib/arm/io.c +1
-0@@ -87,6 +87,34 @@ void puts(const char *s)spin_unlock (&uart_lock); } +static int ____getchar (void )
Three underscores? ... that's quite a lot already. I'd maybe rather name the function "do_getchar" or something similar instead. Or simply merge the code into the __getchar function below - it's just three lines.
lib/arm/io.c void puts(const char *s) · line 98
Quoted 8 lines
lib/arm/io.c +8
-0+{ +int c; + +spin_lock (&uart_lock); + c =readb (uart0_base); +spin_unlock (&uart_lock); + +return c ? :-1 ;
Just a matter of taste, but I prefer the elvis operator without space in between.
lib/arm/io.c void puts(const char *s) · line 104
Quoted 6 lines
lib/arm/io.c +6
-0+} + +/* +* Minimalist implementation for migration completion detection. +* Without FIFOs enabled on the QEMU UART device we just read +* the data register: we cannot read more than 16 characters.
Where are the 16 bytes buffered if FIFOs are disabled?
lib/arm/io.c void puts(const char *s) · line 117
Quoted 13 lines
lib/arm/io.c +13
-0+*/ +int __getchar (void ) +{ +int c =____getchar (); +static int count; + +if (c !=-1 ) + ++count; + +assert (count <16 ); + +return c; +}
The above comments were only nits ... feel free to ignore them if you don't want to respin the series just because of this.
Thomas
Re: [kvm-unit-tests PATCH v3 12/14] arm/run: Allow Migration tests
Hi Thomas,
Quoted 62 lines · Thomas Huth
Quoted 61 lines · Eric Auger
Let's link getchar.o to use puts and getchar from the tests.
Then allow tests belonging to the migration group to trigger the migration from the test code by putting "migrate" into the uart. Then the code can wait for the migration completion by using
getchar().The
__getcharimplement is minimalist as it just reads the data register. It is just meant to read the single character emitted at the end of the migration by the runner script.It is not meant to read more data (FIFOs are not enabled).
Signed-off-by Eric Auger <[email protected]>---
v2 -> v3:
- take the lock
- assert if more than 16 chars
- removed Thomas' R-b--- arm/Makefile.common | 2 +- arm/run | 2 +- lib/arm/io.c | 28 ++++++++++++++++++++++++++++ 3 files changed, 30 insertions(+), 2 deletions(-)arm/Makefile.common +1
-1@@ -32,7 +32,7 @@ CFLAGS += -I $(SRCDIR)/lib -I $(SRCDIR)/lib/libfdt -I lib asm-offsets = lib/$(ARCH)/asm-offsets.h include $(SRCDIR)/scripts/asm-offsets.mak -cflatobjs += lib/util.o +cflatobjs += lib/util.o lib/getchar.o cflatobjs += lib/alloc_phys.o cflatobjs += lib/alloc_page.o cflatobjs += lib/vmalloc.oarm/run +1
-1@@ -61,6 +61,6 @@ fi M+=",accel=$ACCEL" command="$qemu -nodefaults $M -cpu $processor $chr_testdev $pci_testdev" command+=" -display none -serial stdio -kernel" -command="$(timeout_cmd) $command" +command="$(migration_cmd) $(timeout_cmd) $command" run_qemu $command "$@"lib/arm/io.c +0
-0@@ -87,6 +87,34 @@ void puts(const char *s)spin_unlock (&uart_lock); }
Quoted 5 lines
lib/arm/io.c void puts(const char *s) · line 90
Quoted 1 line
lib/arm/io.c +1
-0+static int ____getchar (void )Three underscores? ... that's quite a lot already. I'd maybe rather name the function "
do_getchar" or something similar instead. Or simply merge the code into the__getcharfunction below - it's just three lines.
OK
Quoted 12 lines
lib/arm/io.c void puts(const char *s) · line 98
Quoted 9 lines
lib/arm/io.c +8
-0+{ +int c; + +spin_lock (&uart_lock); + c =readb (uart0_base); +spin_unlock (&uart_lock); + +return c ? :-1 ;Just a matter of taste, but I prefer the elvis operator without space in between.
OK
Quoted 9 lines
lib/arm/io.c void puts(const char *s) · line 104
Quoted 7 lines
lib/arm/io.c +6
-0+} + +/* +* Minimalist implementation for migration completion detection. +* Without FIFOs enabled on the QEMU UART device we just read +* the data register: we cannot read more than 16 characters. Where are the 16 bytes buffered if FIFOs are disabled?
I think this is in the PL011 data register (UARTDR), 4 words.
https://developer.arm.com/docs/ddi0183/latest/programmers-model/register-descriptions/data-register-uartdr
Quoted 17 lines
Quoted 4 lines
lib/arm/io.c +3
-0+*/ +int __getchar (void ) +{lib/arm/io.c void puts(const char *s) · line 117
Quoted 10 lines
lib/arm/io.c +10
-0+int c =____getchar (); +static int count; + +if (c !=-1 ) + ++count; + +assert (count <16 ); + +return c; +}The above comments were only nits ... feel free to ignore them if you don't want to respin the series just because of this.
No Problem. Thank you for your time.
Thanks
Eric
Quoted 3 lines
Thomas
PATCH v3 13/14 arm/arm64: ITS: migration tests
In reply to Eric AugerThis test maps LPIs (populates the device table, the collection table, interrupt translation tables, configuration table), migrates and make sure the translation is correct on the destination.
--- arm/gic.c | 59 ++++++++++++++++++++++++++++++++++++---- arm/unittests.cfg | 8 ++++++ lib/arm/asm/gic-v3-its.h | 2 ++ lib/arm/gic-v3-its.c | 22 +++++++++++++++ 4 files changed, 85 insertions(+), 6 deletions(-)
arm/gic.c +53 -6
@@ -593,6 +593,7 @@ static void gic_test_mmio(void)static void test_its_introspection (void ) {}static void test_its_trigger (void ) {} +static void test_its_migration (void ) {}#else /* __arch64__ */
@@ -665,13 +666,19 @@ static bool its_prerequisites(int nb_cpus)return false; } -static void test_its_trigger (void ) +/* +* Setup the configuration for those mappings: +* dev_id=2 event=20 -> vcpu 3, intid=8195 +* dev_id=7 event=255 -> vcpu 2, intid=8196 +* LPIs ready to hit +*/ +static int its_setup1 (void ) {struct its_collection *col3, *col2;struct its_device *dev2, *dev7;if (its_prerequisites (4 )) -return ; +return -1 ; dev2 =its_create_device (2 /* dev id */ ,8 /* nb_ites */ ); dev7 =its_create_device (7 /* dev id */ ,8 /* nb_ites */ );
@@ -685,14 +692,10 @@ static void test_its_trigger(void)its_send_invall (col2);its_send_invall (col3); -report_prefix_push ("int" );/* * dev=2, eventid=20 -> lpi= 8195, col=3 * dev=7, eventid=255 -> lpi= 8196, col=2 -* Trigger dev2, eventid=20 and dev7, eventid=255 -* Check both LPIs hit */ -its_send_mapd (dev2, true);its_send_mapd (dev7, true);
@@ -703,6 +706,23 @@ static void test_its_trigger(void) 20 /* event id */, col3);its_send_mapti (dev7,8196 /* lpi id */ ,255 /* event id */ , col2); +return 0 ; +} + +static void test_its_trigger (void ) +{ +struct its_collection *col3, *col2; +struct its_device *dev2, *dev7; + +if (its_setup1 ()) +return ; + + col3 =its_get_collection (3 ); + col2 =its_get_collection (2 ); + dev2 =its_get_device (2 ); + dev7 =its_get_device (7 ); + +report_prefix_push ("int" );lpi_stats_expect (3 ,8195 );its_send_int (dev2,20 );
@@ -763,6 +783,29 @@ static void test_its_trigger(void)check_lpi_stats (); report_prefix_pop(); } + +static void test_its_migration (void ) +{ +struct its_device *dev2, *dev7; + +if (its_setup1 ()) +return ; + + dev2 =its_get_device (2 ); + dev7 =its_get_device (7 ); + +puts ("Now migrate the VM, then press a key to continue...\n" ); + (void )getchar (); +report (true,"Migration complete" ); + +lpi_stats_expect (3 ,8195 ); +its_send_int (dev2,20 ); +check_lpi_stats (); + +lpi_stats_expect (2 ,8196 ); +its_send_int (dev7,255 ); +check_lpi_stats (); +}#endif int main (int argc,char **argv)
@@ -800,6 +843,10 @@ int main(int argc, char **argv)report_prefix_push (argv[1 ]);test_its_trigger ();report_prefix_pop (); + }else if (!strcmp (argv[1 ],"its-migration" )) { +report_prefix_push (argv[1 ]); +test_its_migration (); +report_prefix_pop (); }else if (strcmp (argv[1 ],"its-introspection" ) ==0 ) {report_prefix_push (argv[1 ]);test_its_introspection ();
arm/unittests.cfg +8 -0
@@ -136,6 +136,14 @@ extra_params = -machine gic-version=3 -append 'its-trigger' groups = its arch = arm64 +[its-migration] +file = gic.flat +smp = $MAX_SMP +accel = kvm +extra_params = -machine gic-version=3 -append 'its-migration' +groups = its migration +arch = arm64 + # Test PSCI emulation [psci] file = psci.flat
lib/arm/asm/gic-v3-its.h +2 -0
@@ -151,6 +151,8 @@ extern void its_send_invall(struct its_collection *col);extern void its_send_movi (struct its_device *dev,struct its_collection *col,u32 id);extern void its_send_sync (struct its_collection *col); +extern struct its_device *its_get_device (u32 id); +extern struct its_collection *its_get_collection (u32 id);#define ITS_FLAGS_CMDQ_NEEDS_FLUSHING (1ULL << 0)#define ITS_FLAGS_WORKAROUND_CAVIUM_22375 (1ULL << 1)
lib/arm/gic-v3-its.c +22 -0
@@ -219,3 +219,25 @@ struct its_collection *its_create_collection(u32 col_id, u32 pe) its_data.nr_collections++;return new; } + +struct its_device *its_get_device (u32 id) +{ +int i; + +for (i =0 ; i <GITS_MAX_DEVICES ; i++) { +if (its_data.devices[i].device_id == id) +return &its_data.devices[i]; + } +return NULL ; +} + +struct its_collection *its_get_collection (u32 id) +{ +int i; + +for (i =0 ; i <GITS_MAX_COLLECTIONS ; i++) { +if (its_data.collections[i].col_id == id) +return &its_data.collections[i]; + } +return NULL ; +}
Signature · 1 line
2.20.1
Re: [kvm-unit-tests PATCH v3 13/14] arm/arm64: ITS: migration tests
Quoted 33 lines · Eric Auger
This test maps LPIs (populates the device table, the collection table, interrupt translation tables, configuration table), migrates and make sure the translation is correct on the destination.
Signed-off-by Eric Auger <[email protected]>--- arm/gic.c | 59 ++++++++++++++++++++++++++++++++++++---- arm/unittests.cfg | 8 ++++++ lib/arm/asm/gic-v3-its.h | 2 ++ lib/arm/gic-v3-its.c | 22 +++++++++++++++ 4 files changed, 85 insertions(+), 6 deletions(-)arm/gic.c +5
-1@@ -593,6 +593,7 @@ static void gic_test_mmio(void)static void test_its_introspection (void ) {}static void test_its_trigger (void ) {} +static void test_its_migration (void ) {}#else /* __arch64__ */ @@ -665,13 +666,19 @@ static bool its_prerequisites(int nb_cpus)return false; } -static void test_its_trigger (void ) +/* +* Setup the configuration for those mappings: +* dev_id=2 event=20 -> vcpu 3, intid=8195 +* dev_id=7 event=255 -> vcpu 2, intid=8196
arm/gic.c static bool its_prerequisites(int nb_cpus) · line 681
Quoted 10 lines
arm/gic.c +4
-1+* LPIs ready to hit +*/ +static int its_setup1 (void ) {struct its_collection *col3, *col2;struct its_device *dev2, *dev7;if (its_prerequisites (4 )) -return ; +return -1 ;
Why not make its_setup1 a bool? Where true means success and false mean failure?
Quoted 50 lines
arm/gic.c +21
-4dev2 =its_create_device (2 /* dev id */ ,8 /* nb_ites */ ); dev7 =its_create_device (7 /* dev id */ ,8 /* nb_ites */ );@@ -685,14 +692,10 @@ static void test_its_trigger(void)its_send_invall (col2);its_send_invall (col3); -report_prefix_push ("int" );/* * dev=2, eventid=20 -> lpi= 8195, col=3 * dev=7, eventid=255 -> lpi= 8196, col=2 -* Trigger dev2, eventid=20 and dev7, eventid=255 -* Check both LPIs hit */ -its_send_mapd (dev2, true);its_send_mapd (dev7, true);@@ -703,6 +706,23 @@ static void test_its_trigger(void) 20 /* event id */, col3);its_send_mapti (dev7,8196 /* lpi id */ ,255 /* event id */ , col2); +return 0 ; +} + +static void test_its_trigger (void ) +{ +struct its_collection *col3, *col2; +struct its_device *dev2, *dev7; + +if (its_setup1 ()) +return ; + + col3 =its_get_collection (3 ); + col2 =its_get_collection (2 ); + dev2 =its_get_device (2 ); + dev7 =its_get_device (7 ); + +report_prefix_push ("int" );lpi_stats_expect (3 ,8195 );its_send_int (dev2,20 );@@ -763,6 +783,29 @@ static void test_its_trigger(void)check_lpi_stats ();report_prefix_pop (); } + +static void test_its_migration (void ) +{ +struct its_device *dev2, *dev7;
arm/gic.c static void test_its_trigger(void) · line 799
Quoted 10 lines
arm/gic.c +10
-0+ +if (its_setup1 ()) +return ; + + dev2 =its_get_device (2 ); + dev7 =its_get_device (7 ); + +puts ("Now migrate the VM, then press a key to continue...\n" ); + (void )getchar (); +report (true,"Migration complete" );
This seems more like a report_info place. If migration fails and we don't complete we'll never get the report FAIL anyway.
Quoted 75 lines
arm/gic.c +13
-0+ +lpi_stats_expect (3 ,8195 ); +its_send_int (dev2,20 ); +check_lpi_stats (); + +lpi_stats_expect (2 ,8196 ); +its_send_int (dev7,255 ); + check_lpi_stats(); +}#endif int main (int argc,char **argv)@@ -800,6 +843,10 @@ int main(int argc, char **argv)report_prefix_push (argv[1 ]);test_its_trigger ();report_prefix_pop (); + }else if (!strcmp (argv[1 ],"its-migration" )) { +report_prefix_push (argv[1 ]); +test_its_migration (); +report_prefix_pop (); }else if (strcmp (argv[1 ],"its-introspection" ) ==0 ) {report_prefix_push (argv[1 ]);test_its_introspection ();arm/unittests.cfg +8
-0@@ -136,6 +136,14 @@ extra_params = -machine gic-version=3 -append 'its-trigger' groups = its arch = arm64 +[its-migration] +file = gic.flat +smp = $MAX_SMP +accel = kvm +extra_params = -machine gic-version=3 -append 'its-migration' +groups = its migration +arch = arm64 + # Test PSCI emulation [psci] file = psci.flatlib/arm/asm/gic-v3-its.h +2
-0@@ -151,6 +151,8 @@ extern void its_send_invall(struct its_collection *col);extern void its_send_movi (struct its_device *dev,struct its_collection *col,u32 id);extern void its_send_sync (struct its_collection *col); +extern struct its_device *its_get_device (u32 id); +extern struct its_collection *its_get_collection (u32 id);#define ITS_FLAGS_CMDQ_NEEDS_FLUSHING (1ULL << 0)#define ITS_FLAGS_WORKAROUND_CAVIUM_22375 (1ULL << 1)lib/arm/gic-v3-its.c +12
-0@@ -219,3 +219,25 @@ struct its_collection *its_create_collection(u32 col_id, u32 pe) its_data.nr_collections++;return new; } + +struct its_device *its_get_device (u32 id) +{ +int i; + +for (i =0 ; i <GITS_MAX_DEVICES ; i++) { +if (its_data.devices[i].device_id == id) +return &its_data.devices[i]; + } +return NULL ; +} +
lib/arm/gic-v3-its.c struct its_collection *its_create_collection(u32 col_id, u32 pe) · line 243
Quoted 10 lines
lib/arm/gic-v3-its.c +10
-0+struct its_collection *its_get_collection (u32 id) +{ +int i; + +for (i =0 ; i <GITS_MAX_COLLECTIONS ; i++) { +if (its_data.collections[i].col_id == id) +return &its_data.collections[i]; + } +return NULL ; +}
The callers aren't checking for NULL. Should we assert here or in the caller?
Thanks,
drew
Quoted 3 lines
Signature · 1 line
2.20.1
Re: [kvm-unit-tests PATCH v3 13/14] arm/arm64: ITS: migration tests
Hi Drew,
Quoted 41 lines · Andrew Jones
Quoted 40 lines · Eric Auger
This test maps LPIs (populates the device table, the collection table, interrupt translation tables, configuration table), migrates and make sure the translation is correct on the destination.
Signed-off-by Eric Auger <[email protected]>--- arm/gic.c | 59 ++++++++++++++++++++++++++++++++++++---- arm/unittests.cfg | 8 ++++++ lib/arm/asm/gic-v3-its.h | 2 ++ lib/arm/gic-v3-its.c | 22 +++++++++++++++ 4 files changed, 85 insertions(+), 6 deletions(-)arm/gic.c +8
-1@@ -593,6 +593,7 @@ static void gic_test_mmio(void)static void test_its_introspection (void ) {}static void test_its_trigger (void ) {} +static void test_its_migration (void ) {}#else /* __arch64__ */ @@ -665,13 +666,19 @@ static bool its_prerequisites(int nb_cpus)return false; } -static void test_its_trigger (void ) +/* +* Setup the configuration for those mappings: +* dev_id=2 event=20 -> vcpu 3, intid=8195 +* dev_id=7 event=255 -> vcpu 2, intid=8196 +* LPIs ready to hit +*/ +static int its_setup1 (void ) {struct its_collection *col3, *col2;struct its_device *dev2, *dev7;
Quoted 6 lines
arm/gic.c static bool its_prerequisites(int nb_cpus) · line 681
Quoted 3 lines
arm/gic.c +1
-1if (its_prerequisites (4 )) -return ; +return -1 ;Why not make
its_setup1a bool? Where true means success and false mean failure?
I tend to prefer the std error return value convention that the bool return value. I aligned its_prerequisites accordingly.
Quoted 56 lines
Quoted 56 lines
arm/gic.c +26
-4dev2 =its_create_device (2 /* dev id */ ,8 /* nb_ites */ ); dev7 =its_create_device (7 /* dev id */ ,8 /* nb_ites */ );@@ -685,14 +692,10 @@ static void test_its_trigger(void)its_send_invall (col2);its_send_invall (col3); -report_prefix_push ("int" );/* * dev=2, eventid=20 -> lpi= 8195, col=3 * dev=7, eventid=255 -> lpi= 8196, col=2 -* Trigger dev2, eventid=20 and dev7, eventid=255 -* Check both LPIs hit */ -its_send_mapd (dev2, true);its_send_mapd (dev7, true);@@ -703,6 +706,23 @@ static void test_its_trigger(void) 20 /* event id */, col3);its_send_mapti (dev7,8196 /* lpi id */ ,255 /* event id */ , col2); +return 0 ; +} + +static void test_its_trigger (void ) +{ +struct its_collection *col3, *col2; +struct its_device *dev2, *dev7; + +if (its_setup1 ()) +return ; + + col3 =its_get_collection (3 ); + col2 =its_get_collection (2 ); + dev2 =its_get_device (2 ); + dev7 =its_get_device (7 ); + +report_prefix_push ("int" );lpi_stats_expect (3 ,8195 );its_send_int (dev2,20 );@@ -763,6 +783,29 @@ static void test_its_trigger(void)check_lpi_stats ();report_prefix_pop (); } + +static void test_its_migration (void ) +{ +struct its_device *dev2, *dev7; + +if (its_setup1 ()) +return ; + + dev2 =its_get_device (2 );
Quoted 8 lines
arm/gic.c static void test_its_trigger(void) · line 799
Quoted 5 lines
arm/gic.c +5
-0+ dev7 =its_get_device (7 ); + +puts ("Now migrate the VM, then press a key to continue...\n" ); + (void )getchar (); +report (true,"Migration complete" );This seems more like a
report_infoplace. If migration fails and we don't complete we'll never get the report FAIL anyway.
OK
Quoted 81 lines
Quoted 81 lines
arm/gic.c +13
-0+ +lpi_stats_expect (3 ,8195 ); +its_send_int (dev2,20 ); +check_lpi_stats (); + +lpi_stats_expect (2 ,8196 ); +its_send_int (dev7,255 ); + check_lpi_stats(); +}#endif int main (int argc,char **argv)@@ -800,6 +843,10 @@ int main(int argc, char **argv)report_prefix_push (argv[1 ]);test_its_trigger ();report_prefix_pop (); + }else if (!strcmp (argv[1 ],"its-migration" )) { +report_prefix_push (argv[1 ]); +test_its_migration (); +report_prefix_pop (); }else if (strcmp (argv[1 ],"its-introspection" ) ==0 ) {report_prefix_push (argv[1 ]);test_its_introspection ();arm/unittests.cfg +8
-0@@ -136,6 +136,14 @@ extra_params = -machine gic-version=3 -append 'its-trigger' groups = its arch = arm64 +[its-migration] +file = gic.flat +smp = $MAX_SMP +accel = kvm +extra_params = -machine gic-version=3 -append 'its-migration' +groups = its migration +arch = arm64 + # Test PSCI emulation [psci] file = psci.flatlib/arm/asm/gic-v3-its.h +2
-0@@ -151,6 +151,8 @@ extern void its_send_invall(struct its_collection *col);extern void its_send_movi (struct its_device *dev,struct its_collection *col,u32 id);extern void its_send_sync (struct its_collection *col); +extern struct its_device *its_get_device (u32 id); +extern struct its_collection *its_get_collection (u32 id);#define ITS_FLAGS_CMDQ_NEEDS_FLUSHING (1ULL << 0)#define ITS_FLAGS_WORKAROUND_CAVIUM_22375 (1ULL << 1)lib/arm/gic-v3-its.c +17
-0@@ -219,3 +219,25 @@ struct its_collection *its_create_collection(u32 col_id, u32 pe) its_data.nr_collections++;return new; } + +struct its_device *its_get_device (u32 id) +{ +int i; + +for (i =0 ; i <GITS_MAX_DEVICES ; i++) { +if (its_data.devices[i].device_id == id) +return &its_data.devices[i]; + } +return NULL ; +} + +struct its_collection *its_get_collection (u32 id) +{ +int i; + +for (i =0 ; i <GITS_MAX_COLLECTIONS ; i++) {
Quoted 8 lines
lib/arm/gic-v3-its.c struct its_collection *its_create_collection(u32 col_id, u32 pe) · line 243
Quoted 5 lines
lib/arm/gic-v3-its.c +5
-0+if (its_data.collections[i].col_id == id) +return &its_data.collections[i]; + } +return NULL ; +}The callers aren't checking for NULL. Should we assert here or in the caller?
Added the assert here.
Thanks
Eric
Quoted 8 lines
Thanks,
drewQuoted 3 lines
Signature · 1 line
2.20.1
PATCH v3 14/14 arm/arm64: ITS: pending table migration test
In reply to Eric AugerAdd two new migration tests. One testing the migration of a topology where collection were unmapped. The second test checks the migration of the pending table.
---
v2 -> v3:
- tests belong to both its and migration groups
--- arm/gic.c | 150 ++++++++++++++++++++++++++++++++++++++++++++++ arm/unittests.cfg | 16 +++++ 2 files changed, 166 insertions(+)
arm/gic.c +150 -0
@@ -195,6 +195,7 @@ static void lpi_handler(struct pt_regs *regs __unused)smp_rmb ();/* pairs with wmb in lpi_stats_expect */ lpi_stats.observed.cpu_id =smp_processor_id (); lpi_stats.observed.lpi_id = irqnr; + acked[lpi_stats.observed.cpu_id]++;smp_wmb ();/* pairs with rmb in check_lpi_stats */ }
@@ -239,6 +240,18 @@ static void secondary_lpi_test(void)while (1 ) wfi(); } + +static void check_lpi_hits (int *expected) +{ +int i; + +for (i =0 ; i < nr_cpus; i++) { +if (acked[i] != expected[i]) +report (false,"expected %d LPIs on PE #%d, %d observed" , + expected[i], i, acked[i]); + } +report (true,"check LPI on all vcpus" ); +}#endif static void gicv2_ipi_send_self (void )
@@ -594,6 +607,8 @@ static void gic_test_mmio(void)static void test_its_introspection (void ) {}static void test_its_trigger (void ) {}static void test_its_migration (void ) {} +static void test_migrate_unmapped_collection (void ) {} +static void test_its_pending_migration (void ) {}#else /* __arch64__ */
@@ -666,6 +681,18 @@ static bool its_prerequisites(int nb_cpus)return false; } +static void set_lpi (struct its_device *dev,u32 eventid,u32 physid, +struct its_collection *col) +{ +if (!dev || !col) +report_abort ("wrong device or collection" ); + +its_send_mapti (dev, physid, eventid, col); + +gicv3_lpi_set_config (physid,LPI_PROP_DEFAULT ); +its_send_invall (col); +} +/* * Setup the configuration for those mappings: * dev_id=2 event=20 -> vcpu 3, intid=8195
@@ -806,6 +833,121 @@ static void test_its_migration(void)its_send_int (dev7,255 );check_lpi_stats (); } + +static void test_migrate_unmapped_collection (void ) +{ +struct its_collection *col; +struct its_device *dev2, *dev7; +u8 config; + +if (its_setup1 ()) +return ; + + col =its_create_collection (nr_cpus -1 , nr_cpus -1 ); + dev2 =its_get_device (2 ); + dev7 =its_get_device (7 ); + +/* MAPTI with the collection unmapped */ +set_lpi (dev2,0 ,8192 , col); + +puts ("Now migrate the VM, then press a key to continue...\n" ); + (void )getchar (); +report (true,"Migration complete" ); + +/* on the destination, map the collection */ +its_send_mapc (col, true); + +lpi_stats_expect (2 ,8196 ); +its_send_int (dev7,255 ); +check_lpi_stats (); + + config =gicv3_lpi_get_config (8192 ); +report (config ==LPI_PROP_DEFAULT , +"Config of LPI 8192 was properly migrated" ); + +lpi_stats_expect (nr_cpus -1 ,8192 ); +its_send_int (dev2,0 ); +check_lpi_stats (); + +/* unmap the collection */ +its_send_mapc (col, false); + +lpi_stats_expect (-1 ,-1 ); +its_send_int (dev2,0 ); +check_lpi_stats (); + +/* remap event 0 onto lpiid 8193 */ +set_lpi (dev2,0 ,8193 , col); +lpi_stats_expect (-1 ,-1 ); +its_send_int (dev2,0 ); +check_lpi_stats (); + +/* remap the collection */ +its_send_mapc (col, true); +lpi_stats_expect (nr_cpus -1 ,8193 ); +} + +static void test_its_pending_migration (void ) +{ +struct its_device *dev; +struct its_collection *collection[2 ]; +int expected[NR_CPUS ]; +u64 pendbaser; +void *ptr; +int i; + +if (its_prerequisites (4 )) +return ; + + dev =its_create_device (2 /* dev id */ ,8 /* nb_ites */ ); +its_send_mapd (dev, true); + + collection[0 ] =its_create_collection (nr_cpus -1 , nr_cpus -1 ); + collection[1 ] =its_create_collection (nr_cpus -2 , nr_cpus -2 ); +its_send_mapc (collection[0 ], true); +its_send_mapc (collection[1 ], true); + +/* disable lpi at redist level */ +gicv3_lpi_rdist_ctrl (nr_cpus -1 , false); +gicv3_lpi_rdist_ctrl (nr_cpus -2 , false); + +/* even lpis are assigned to even cpu */ +for (i =0 ; i <256 ; i++) { +struct its_collection *col = i %2 ? collection[0 ] : + collection[1 ]; +int vcpu = col->target_address >>16 ; + +its_send_mapti (dev,8192 + i, i, col); +gicv3_lpi_set_config (8192 + i,LPI_PROP_DEFAULT ); +gicv3_lpi_set_pending_table_bit (vcpu,8192 + i, true); + } +its_send_invall (collection[0 ]); +its_send_invall (collection[1 ]); + +/* Set the PTZ bit on each pendbaser */ + + expected[nr_cpus -1 ] =128 ; + expected[nr_cpus -2 ] =128 ; + + ptr = gicv3_data.redist_base[nr_cpus -1 ] +GICR_PENDBASER ; + pendbaser =readq (ptr); +writeq (pendbaser & ~GICR_PENDBASER_PTZ , ptr); + + ptr = gicv3_data.redist_base[nr_cpus -2 ] +GICR_PENDBASER ; + pendbaser =readq (ptr); +writeq (pendbaser & ~GICR_PENDBASER_PTZ , ptr); + +gicv3_lpi_rdist_ctrl (nr_cpus -1 , true); +gicv3_lpi_rdist_ctrl (nr_cpus -2 , true); + +puts ("Now migrate the VM, then press a key to continue...\n" ); + (void )getchar (); +report (true,"Migration complete" ); + +mdelay (1000 ); + +check_lpi_hits (expected); +}#endif int main (int argc,char **argv)
@@ -847,6 +989,14 @@ int main(int argc, char **argv)report_prefix_push (argv[1 ]);test_its_migration ();report_prefix_pop (); + }else if (!strcmp (argv[1 ],"its-pending-migration" )) { +report_prefix_push (argv[1 ]); +test_its_pending_migration (); +report_prefix_pop (); + }else if (!strcmp (argv[1 ],"its-migrate-unmapped-collection" )) { +report_prefix_push (argv[1 ]); +test_migrate_unmapped_collection (); +report_prefix_pop (); }else if (strcmp (argv[1 ],"its-introspection" ) ==0 ) {report_prefix_push (argv[1 ]);test_its_introspection ();
arm/unittests.cfg +16 -0
@@ -144,6 +144,22 @@ extra_params = -machine gic-version=3 -append 'its-migration' groups = its migration arch = arm64 +[its-pending-migration] +file = gic.flat +smp = $MAX_SMP +accel = kvm +extra_params = -machine gic-version=3 -append 'its-pending-migration' +groups = its migration +arch = arm64 + +[its-migrate-unmapped-collection] +file = gic.flat +smp = $MAX_SMP +accel = kvm +extra_params = -machine gic-version=3 -append 'its-migrate-unmapped-collection' +groups = its migration +arch = arm64 + # Test PSCI emulation [psci] file = psci.flat
Signature · 1 line
2.20.1
Re: [kvm-unit-tests PATCH v3 14/14] arm/arm64: ITS: pending table migration test
Quoted 30 lines · Eric Auger
Add two new migration tests. One testing the migration of a topology where collection were unmapped. The second test checks the migration of the pending table.
Signed-off-by Eric Auger <[email protected]>---
v2 -> v3:
- tests belong to both its and migration groups--- arm/gic.c | 150 ++++++++++++++++++++++++++++++++++++++++++++++ arm/unittests.cfg | 16 +++++ 2 files changed, 166 insertions(+)arm/gic.c +1
-0@@ -195,6 +195,7 @@ static void lpi_handler(struct pt_regs *regs __unused)smp_rmb ();/* pairs with wmb in lpi_stats_expect */ lpi_stats.observed.cpu_id =smp_processor_id (); lpi_stats.observed.lpi_id = irqnr; + acked[lpi_stats.observed.cpu_id]++;smp_wmb ();/* pairs with rmb in check_lpi_stats */ }@@ -239,6 +240,18 @@ static void secondary_lpi_test(void)while (1 )wfi ();
arm/gic.c static void secondary_lpi_test(void) · line 251
Quoted 10 lines
arm/gic.c +9
-0} + +static void check_lpi_hits (int *expected) +{ +int i; + +for (i =0 ; i < nr_cpus; i++) { +if (acked[i] != expected[i]) +report (false,"expected %d LPIs on PE #%d, %d observed" , + expected[i], i, acked[i]);
report_info
pass = false
arm/gic.c static void secondary_lpi_test(void) · line 253
Quoted 2 lines
arm/gic.c +2
-0+ } +report (true,"check LPI on all vcpus" );
report(pass, ...);
arm/gic.c static void gic_test_mmio(void) · line 611
Quoted 10 lines
arm/gic.c +3
-0+}#endif static void gicv2_ipi_send_self (void )@@ -594,6 +607,8 @@ static void gic_test_mmio(void)static void test_its_introspection (void ) {}static void test_its_trigger (void ) {}static void test_its_migration (void ) {} +static void test_migrate_unmapped_collection (void ) {} +static void test_its_pending_migration (void ) {}
I'm not sure what's worse. This pile of stubs or one #ifdef in main() wrapping all the calls.
arm/gic.c static bool its_prerequisites(int nb_cpus) · line 687
Quoted 11 lines
arm/gic.c +4
-0#else /* __arch64__ */ @@ -666,6 +681,18 @@ static bool its_prerequisites(int nb_cpus)return false; } +static void set_lpi (struct its_device *dev,u32 eventid,u32 physid, +struct its_collection *col) +{ +if (!dev || !col)
I don't think col can be null, and this doesn't look like the right place to check if dev is null. If we're bothiner to call set_lpi, then I think we should already expect dev to be good to go.
Quoted 25 lines
arm/gic.c +18
-0+report_abort ("wrong device or collection" ); + +its_send_mapti (dev, physid, eventid, col); + +gicv3_lpi_set_config (physid,LPI_PROP_DEFAULT ); +its_send_invall (col); +} +/* * Setup the configuration for those mappings: * dev_id=2 event=20 -> vcpu 3, intid=8195 @@ -806,6 +833,121 @@ static void test_its_migration(void)its_send_int (dev7,255 );check_lpi_stats (); } + +static void test_migrate_unmapped_collection (void ) +{ +struct its_collection *col; +struct its_device *dev2, *dev7; +u8 config; + +if (its_setup1 ()) +return ; +
arm/gic.c static void test_its_migration(void) · line 855
Quoted 10 lines
arm/gic.c +10
-0+ col =its_create_collection (nr_cpus -1 , nr_cpus -1 ); + dev2 =its_get_device (2 ); + dev7 =its_get_device (7 ); + +/* MAPTI with the collection unmapped */ +set_lpi (dev2,0 ,8192 , col); + +puts ("Now migrate the VM, then press a key to continue...\n" ); + (void )getchar (); +report (true,"Migration complete" );
report_info
Quoted 29 lines
arm/gic.c +29
-0+ +/* on the destination, map the collection */ +its_send_mapc (col, true); + +lpi_stats_expect (2 ,8196 ); +its_send_int (dev7,255 ); +check_lpi_stats (); + + config =gicv3_lpi_get_config (8192 ); +report (config ==LPI_PROP_DEFAULT , +"Config of LPI 8192 was properly migrated" ); + +lpi_stats_expect (nr_cpus -1 ,8192 ); +its_send_int (dev2,0 ); +check_lpi_stats (); + +/* unmap the collection */ +its_send_mapc (col, false); + +lpi_stats_expect (-1 ,-1 ); +its_send_int (dev2,0 ); +check_lpi_stats (); + +/* remap event 0 onto lpiid 8193 */ +set_lpi (dev2,0 ,8193 , col); +lpi_stats_expect (-1 ,-1 ); +its_send_int (dev2,0 ); +check_lpi_stats (); +
arm/gic.c static void test_its_migration(void) · line 894
Quoted 10 lines
arm/gic.c +10
-0+/* remap the collection */ +its_send_mapc (col, true); +lpi_stats_expect (nr_cpus -1 ,8193 ); +} + +static void test_its_pending_migration (void ) +{ +struct its_device *dev; +struct its_collection *collection[2 ]; +int expected[NR_CPUS ];
expected = malloc(nr_cpus * sizeof(int));
I know there are other places using NR_CPUS right now that don't have to, but we shouldn't add more. Eventually I'll change the other places too.
Quoted 14 lines
arm/gic.c +14
-0+u64 pendbaser; +void *ptr; +int i; + +if (its_prerequisites (4 )) +return ; + + dev =its_create_device (2 /* dev id */ ,8 /* nb_ites */ ); +its_send_mapd (dev, true); + + collection[0 ] =its_create_collection (nr_cpus -1 , nr_cpus -1 ); + collection[1 ] =its_create_collection (nr_cpus -2 , nr_cpus -2 ); +its_send_mapc (collection[0 ], true); +its_send_mapc (collection[1 ], true);
arm/gic.c static void test_its_migration(void) · line 918
Quoted 10 lines
arm/gic.c +10
-0+ +/* disable lpi at redist level */ +gicv3_lpi_rdist_ctrl (nr_cpus -1 , false); +gicv3_lpi_rdist_ctrl (nr_cpus -2 , false); + +/* even lpis are assigned to even cpu */ +for (i =0 ; i <256 ; i++) { +struct its_collection *col = i %2 ? collection[0 ] : + collection[1 ]; +int vcpu = col->target_address >>16 ;
I'm lost with the even/odd (nr_cpus - 1)/(nr_cpus - 2) stuff, and won't it swap if nr_cpus is odd vs. even?
Shouldn't we just have something like
pe1 = nr_cpus - 1; pe2 = nr_cpus - 2; col1 = its_create_collection(pe1, pe1); col2 = its_create_collection(pe2, pe2);
without mentioning even and odd?
Quoted 17 lines
arm/gic.c +17
-0+ +its_send_mapti (dev,8192 + i, i, col); +gicv3_lpi_set_config (8192 + i,LPI_PROP_DEFAULT ); +gicv3_lpi_set_pending_table_bit (vcpu,8192 + i, true); + } +its_send_invall (collection[0 ]); +its_send_invall (collection[1 ]); + +/* Set the PTZ bit on each pendbaser */ + + expected[nr_cpus -1 ] =128 ; + expected[nr_cpus -2 ] =128 ; + + ptr = gicv3_data.redist_base[nr_cpus -1 ] +GICR_PENDBASER ; + pendbaser =readq (ptr); +writeq (pendbaser & ~GICR_PENDBASER_PTZ , ptr); +
arm/gic.c static void test_its_migration(void) · line 945
Quoted 10 lines
arm/gic.c +10
-0+ ptr = gicv3_data.redist_base[nr_cpus -2 ] +GICR_PENDBASER ; + pendbaser =readq (ptr); +writeq (pendbaser & ~GICR_PENDBASER_PTZ , ptr); + +gicv3_lpi_rdist_ctrl (nr_cpus -1 , true); +gicv3_lpi_rdist_ctrl (nr_cpus -2 , true); + +puts ("Now migrate the VM, then press a key to continue...\n" ); + (void )getchar (); +report (true,"Migration complete" );
report_info
arm/gic.c static void test_its_migration(void) · line 947
Quoted 2 lines
arm/gic.c +2
-0+ +mdelay (1000 );
This delay needs a comment explaining why it's here.
Quoted 52 lines
arm/gic.c +11
-0+ + check_lpi_hits(expected); +}#endif int main (int argc,char **argv)@@ -847,6 +989,14 @@ int main(int argc, char **argv)report_prefix_push (argv[1 ]);test_its_migration ();report_prefix_pop (); + }else if (!strcmp (argv[1 ],"its-pending-migration" )) { +report_prefix_push (argv[1 ]); +test_its_pending_migration (); +report_prefix_pop (); + }else if (!strcmp (argv[1 ],"its-migrate-unmapped-collection" )) { +report_prefix_push (argv[1 ]); +test_migrate_unmapped_collection (); +report_prefix_pop (); }else if (strcmp (argv[1 ],"its-introspection" ) ==0 ) {report_prefix_push (argv[1 ]);test_its_introspection ();arm/unittests.cfg +16
-0@@ -144,6 +144,22 @@ extra_params = -machine gic-version=3 -append 'its-migration' groups = its migration arch = arm64 +[its-pending-migration] +file = gic.flat +smp = $MAX_SMP +accel = kvm +extra_params = -machine gic-version=3 -append 'its-pending-migration' +groups = its migration +arch = arm64 + +[its-migrate-unmapped-collection] +file = gic.flat +smp = $MAX_SMP +accel = kvm +extra_params = -machine gic-version=3 -append 'its-migrate-unmapped-collection' +groups = its migration +arch = arm64 + # Test PSCI emulation [psci] file = psci.flatSignature · 1 line
2.20.1
Thanks,
drew
_______________________________________________
kvmarm mailing list
[email protected]
https://lists.cs.columbia.edu/mailman/listinfo/kvmarm
Re: [kvm-unit-tests PATCH v3 14/14] arm/arm64: ITS: pending table migration test
Hi Drew,
Quoted 45 lines · Andrew Jones
Quoted 30 lines · Eric Auger
Add two new migration tests. One testing the migration of a topology where collection were unmapped. The second test checks the migration of the pending table.
Signed-off-by Eric Auger <[email protected]>---
v2 -> v3:
- tests belong to both its and migration groups--- arm/gic.c | 150 ++++++++++++++++++++++++++++++++++++++++++++++ arm/unittests.cfg | 16 +++++ 2 files changed, 166 insertions(+)arm/gic.c +1
-0@@ -195,6 +195,7 @@ static void lpi_handler(struct pt_regs *regs __unused)smp_rmb ();/* pairs with wmb in lpi_stats_expect */ lpi_stats.observed.cpu_id =smp_processor_id (); lpi_stats.observed.lpi_id = irqnr; + acked[lpi_stats.observed.cpu_id]++;smp_wmb ();/* pairs with rmb in check_lpi_stats */ }@@ -239,6 +240,18 @@ static void secondary_lpi_test(void)while (1 )wfi ();arm/gic.c static void secondary_lpi_test(void) · line 251
Quoted 10 lines
arm/gic.c +9
-0} + +static void check_lpi_hits (int *expected) +{ +int i; + +for (i =0 ; i < nr_cpus; i++) { +if (acked[i] != expected[i]) +report (false,"expected %d LPIs on PE #%d, %d observed" , + expected[i], i, acked[i]);
report_info
pass = false
Quoted 4 lines
arm/gic.c static void secondary_lpi_test(void) · line 253
Quoted 2 lines
arm/gic.c +2
-0+ } +report (true,"check LPI on all vcpus" );
report(pass, ...);
I still don't get the issue.
Quoted 4 lines
Quoted 4 lines
arm/gic.c +1
-0+}#endif
Quoted 10 lines
arm/gic.c static void gic_test_mmio(void) · line 611
Quoted 7 lines
arm/gic.c +2
-0static void gicv2_ipi_send_self (void )@@ -594,6 +607,8 @@ static void gic_test_mmio(void)static void test_its_introspection (void ) {}static void test_its_trigger (void ) {}static void test_its_migration (void ) {} +static void test_migrate_unmapped_collection (void ) {} +static void test_its_pending_migration (void ) {}I'm not sure what's worse. This pile of stubs or one #ifdef in
main()wrapping all the calls.
Those stubs now are in the arm header.
Quoted 8 lines
Quoted 8 lines
arm/gic.c +0
-0#else /* __arch64__ */ @@ -666,6 +681,18 @@ static bool its_prerequisites(int nb_cpus)return false; }
Quoted 8 lines
arm/gic.c static bool its_prerequisites(int nb_cpus) · line 687
Quoted 4 lines
arm/gic.c +4
-0+static void set_lpi (struct its_device *dev,u32 eventid,u32 physid, +struct its_collection *col) +{ +if (!dev || !col)I don't think col can be null, and this doesn't look like the right place to check if dev is null. If we're bothiner to call
set_lpi, then I think we should already expect dev to be good to go.
put an assert() instead
Quoted 30 lines
Quoted 30 lines
arm/gic.c +22
-0+report_abort ("wrong device or collection" ); + +its_send_mapti (dev, physid, eventid, col); + +gicv3_lpi_set_config (physid,LPI_PROP_DEFAULT ); +its_send_invall (col); +} +/* * Setup the configuration for those mappings: * dev_id=2 event=20 -> vcpu 3, intid=8195 @@ -806,6 +833,121 @@ static void test_its_migration(void)its_send_int (dev7,255 );check_lpi_stats (); } + +static void test_migrate_unmapped_collection (void ) +{ +struct its_collection *col; +struct its_device *dev2, *dev7; +u8 config; + +if (its_setup1 ()) +return ; + + col =its_create_collection (nr_cpus -1 , nr_cpus -1 ); + dev2 =its_get_device (2 ); + dev7 =its_get_device (7 ); +
Quoted 8 lines
arm/gic.c static void test_its_migration(void) · line 855
Quoted 6 lines
arm/gic.c +6
-0+/* MAPTI with the collection unmapped */ +set_lpi (dev2,0 ,8192 , col); + +puts ("Now migrate the VM, then press a key to continue...\n" ); + (void )getchar (); +report (true,"Migration complete" );
report_info
yep
Quoted 43 lines
Quoted 30 lines
arm/gic.c +29
-0+ +/* on the destination, map the collection */ +its_send_mapc (col, true); + +lpi_stats_expect (2 ,8196 ); +its_send_int (dev7,255 ); +check_lpi_stats (); + + config =gicv3_lpi_get_config (8192 ); +report (config ==LPI_PROP_DEFAULT , +"Config of LPI 8192 was properly migrated" ); + +lpi_stats_expect (nr_cpus -1 ,8192 ); +its_send_int (dev2,0 ); +check_lpi_stats (); + +/* unmap the collection */ +its_send_mapc (col, false); + +lpi_stats_expect (-1 ,-1 ); +its_send_int (dev2,0 ); +check_lpi_stats (); + +/* remap event 0 onto lpiid 8193 */ +set_lpi (dev2,0 ,8193 , col); +lpi_stats_expect (-1 ,-1 ); +its_send_int (dev2,0 ); +check_lpi_stats (); +arm/gic.c static void test_its_migration(void) · line 894
Quoted 10 lines
arm/gic.c +10
-0+/* remap the collection */ +its_send_mapc (col, true); +lpi_stats_expect (nr_cpus -1 ,8193 ); +} + +static void test_its_pending_migration (void ) +{ +struct its_device *dev; +struct its_collection *collection[2 ]; +int expected[NR_CPUS ];expected =
malloc(nr_cpus * sizeof(int));
Quoted 2 lines
I know there are other places using
NR_CPUSright now that don't have to, but we shouldn't add more. Eventually I'll change the other places too.
OK
Quoted 29 lines
Quoted 15 lines
arm/gic.c +14
-0+u64 pendbaser; +void *ptr; +int i; + +if (its_prerequisites (4 )) +return ; + + dev =its_create_device (2 /* dev id */ ,8 /* nb_ites */ ); +its_send_mapd (dev, true); + + collection[0 ] =its_create_collection (nr_cpus -1 , nr_cpus -1 ); + collection[1 ] =its_create_collection (nr_cpus -2 , nr_cpus -2 ); +its_send_mapc (collection[0 ], true); +its_send_mapc (collection[1 ], true);arm/gic.c static void test_its_migration(void) · line 918
Quoted 10 lines
arm/gic.c +10
-0+ +/* disable lpi at redist level */ +gicv3_lpi_rdist_ctrl (nr_cpus -1 , false); +gicv3_lpi_rdist_ctrl (nr_cpus -2 , false); + +/* even lpis are assigned to even cpu */ +for (i =0 ; i <256 ; i++) { +struct its_collection *col = i %2 ? collection[0 ] : + collection[1 ]; +int vcpu = col->target_address >>16 ;I'm lost with the even/odd (
nr_cpus- 1)/(nr_cpus - 2) stuff, and won't it swap ifnr_cpusis odd vs. even?
Quoted 8 lines
Shouldn't we just have something like
pe1 = nr_cpus - 1; pe2 = nr_cpus - 2; col1 = its_create_collection(pe1, pe1); col2 = its_create_collection(pe2, pe2);without mentioning even and odd?
OK
Quoted 22 lines
Quoted 22 lines
arm/gic.c +21
-0+ +its_send_mapti (dev,8192 + i, i, col); +gicv3_lpi_set_config (8192 + i,LPI_PROP_DEFAULT ); +gicv3_lpi_set_pending_table_bit (vcpu,8192 + i, true); + } +its_send_invall (collection[0 ]); +its_send_invall (collection[1 ]); + +/* Set the PTZ bit on each pendbaser */ + + expected[nr_cpus -1 ] =128 ; + expected[nr_cpus -2 ] =128 ; + + ptr = gicv3_data.redist_base[nr_cpus -1 ] +GICR_PENDBASER ; + pendbaser =readq (ptr); +writeq (pendbaser & ~GICR_PENDBASER_PTZ , ptr); + + ptr = gicv3_data.redist_base[nr_cpus -2 ] +GICR_PENDBASER ; + pendbaser =readq (ptr); +writeq (pendbaser & ~GICR_PENDBASER_PTZ , ptr); +
Quoted 8 lines
arm/gic.c static void test_its_migration(void) · line 945
Quoted 6 lines
arm/gic.c +6
-0+gicv3_lpi_rdist_ctrl (nr_cpus -1 , true); +gicv3_lpi_rdist_ctrl (nr_cpus -2 , true); + +puts ("Now migrate the VM, then press a key to continue...\n" ); + (void )getchar (); +report (true,"Migration complete" );
report_info
OK
Quoted 5 lines
arm/gic.c static void test_its_migration(void) · line 947
Quoted 3 lines
arm/gic.c +2
-0+ +mdelay (1000 );This delay needs a comment explaining why it's here.
OK
Thanks
Eric
Quoted 57 lines
Quoted 53 lines
arm/gic.c +11
-0+ + check_lpi_hits(expected); +}#endif int main (int argc,char **argv)@@ -847,6 +989,14 @@ int main(int argc, char **argv)report_prefix_push (argv[1 ]);test_its_migration ();report_prefix_pop (); + }else if (!strcmp (argv[1 ],"its-pending-migration" )) { +report_prefix_push (argv[1 ]); +test_its_pending_migration (); +report_prefix_pop (); + }else if (!strcmp (argv[1 ],"its-migrate-unmapped-collection" )) { +report_prefix_push (argv[1 ]); +test_migrate_unmapped_collection (); +report_prefix_pop (); }else if (strcmp (argv[1 ],"its-introspection" ) ==0 ) {report_prefix_push (argv[1 ]);test_its_introspection ();arm/unittests.cfg +16
-0@@ -144,6 +144,22 @@ extra_params = -machine gic-version=3 -append 'its-migration' groups = its migration arch = arm64 +[its-pending-migration] +file = gic.flat +smp = $MAX_SMP +accel = kvm +extra_params = -machine gic-version=3 -append 'its-pending-migration' +groups = its migration +arch = arm64 + +[its-migrate-unmapped-collection] +file = gic.flat +smp = $MAX_SMP +accel = kvm +extra_params = -machine gic-version=3 -append 'its-migrate-unmapped-collection' +groups = its migration +arch = arm64 + # Test PSCI emulation [psci] file = psci.flatSignature · 1 line
2.20.1Thanks,
drew
Re: [kvm-unit-tests PATCH v3 14/14] arm/arm64: ITS: pending table migration test
Quoted 48 lines · Auger Eric
Hi Drew,
Quoted 45 lines · Andrew Jones
Quoted 30 lines · Eric Auger
Add two new migration tests. One testing the migration of a topology where collection were unmapped. The second test checks the migration of the pending table.
Signed-off-by Eric Auger <[email protected]>---
v2 -> v3:
- tests belong to both its and migration groups--- arm/gic.c | 150 ++++++++++++++++++++++++++++++++++++++++++++++ arm/unittests.cfg | 16 +++++ 2 files changed, 166 insertions(+)arm/gic.c +1
-0@@ -195,6 +195,7 @@ static void lpi_handler(struct pt_regs *regs __unused)smp_rmb ();/* pairs with wmb in lpi_stats_expect */ lpi_stats.observed.cpu_id =smp_processor_id (); lpi_stats.observed.lpi_id = irqnr; + acked[lpi_stats.observed.cpu_id]++;smp_wmb ();/* pairs with rmb in check_lpi_stats */ }@@ -239,6 +240,18 @@ static void secondary_lpi_test(void)while (1 )wfi ();arm/gic.c static void secondary_lpi_test(void) · line 251
Quoted 10 lines
arm/gic.c +9
-0} + +static void check_lpi_hits (int *expected) +{ +int i; + +for (i =0 ; i < nr_cpus; i++) { +if (acked[i] != expected[i]) +report (false,"expected %d LPIs on PE #%d, %d observed" , + expected[i], i, acked[i]);
report_info
pass = false
Quoted 5 lines
Quoted 4 lines
arm/gic.c static void secondary_lpi_test(void) · line 253
Quoted 2 lines
arm/gic.c +2
-0+ } +report (true,"check LPI on all vcpus" );
report(pass, ...);I still don't get the issue.
Your messages aren't consistent and have dynamic info. You need
report_info(...); // dynamic info
if (failure_condition) {
pass_boolean = false;
}report(pass_boolean, single_message_for_both_pass_and_failure);
Quoted 255 lines
Quoted 4 lines
Quoted 4 lines
arm/gic.c +1
-0+}#endif Quoted 10 lines
arm/gic.c static void gic_test_mmio(void) · line 611
Quoted 7 lines
arm/gic.c +2
-0static void gicv2_ipi_send_self (void )@@ -594,6 +607,8 @@ static void gic_test_mmio(void)static void test_its_introspection (void ) {}static void test_its_trigger (void ) {}static void test_its_migration (void ) {} +static void test_migrate_unmapped_collection (void ) {} +static void test_its_pending_migration (void ) {}I'm not sure what's worse. This pile of stubs or one #ifdef in
main()wrapping all the calls.Those stubs now are in the arm header.
Quoted 8 lines
Quoted 8 lines
arm/gic.c +0
-0#else /* __arch64__ */ @@ -666,6 +681,18 @@ static bool its_prerequisites(int nb_cpus)return false; }Quoted 8 lines
arm/gic.c static bool its_prerequisites(int nb_cpus) · line 687
Quoted 4 lines
arm/gic.c +4
-0+static void set_lpi (struct its_device *dev,u32 eventid,u32 physid, +struct its_collection *col) +{ +if (!dev || !col)I don't think col can be null, and this doesn't look like the right place to check if dev is null. If we're bothiner to call
set_lpi, then I think we should already expect dev to be good to go.put an
assert()insteadQuoted 30 lines
Quoted 30 lines
arm/gic.c +22
-0+report_abort ("wrong device or collection" ); + +its_send_mapti (dev, physid, eventid, col); + +gicv3_lpi_set_config (physid,LPI_PROP_DEFAULT ); +its_send_invall (col); +} +/* * Setup the configuration for those mappings: * dev_id=2 event=20 -> vcpu 3, intid=8195 @@ -806,6 +833,121 @@ static void test_its_migration(void)its_send_int (dev7,255 );check_lpi_stats (); } + +static void test_migrate_unmapped_collection (void ) +{ +struct its_collection *col; +struct its_device *dev2, *dev7; +u8 config; + +if (its_setup1 ()) +return ; + + col =its_create_collection (nr_cpus -1 , nr_cpus -1 ); + dev2 =its_get_device (2 ); + dev7 =its_get_device (7 ); +Quoted 8 lines
arm/gic.c static void test_its_migration(void) · line 855
Quoted 6 lines
arm/gic.c +6
-0+/* MAPTI with the collection unmapped */ +set_lpi (dev2,0 ,8192 , col); + +puts ("Now migrate the VM, then press a key to continue...\n" ); + (void )getchar (); +report (true,"Migration complete" );
report_infoyep
Quoted 43 lines
Quoted 30 lines
arm/gic.c +29
-0+ +/* on the destination, map the collection */ +its_send_mapc (col, true); + +lpi_stats_expect (2 ,8196 ); +its_send_int (dev7,255 ); +check_lpi_stats (); + + config =gicv3_lpi_get_config (8192 ); +report (config ==LPI_PROP_DEFAULT , +"Config of LPI 8192 was properly migrated" ); + +lpi_stats_expect (nr_cpus -1 ,8192 ); +its_send_int (dev2,0 ); +check_lpi_stats (); + +/* unmap the collection */ +its_send_mapc (col, false); + +lpi_stats_expect (-1 ,-1 ); +its_send_int (dev2,0 ); +check_lpi_stats (); + +/* remap event 0 onto lpiid 8193 */ +set_lpi (dev2,0 ,8193 , col); +lpi_stats_expect (-1 ,-1 ); +its_send_int (dev2,0 ); +check_lpi_stats (); +arm/gic.c static void test_its_migration(void) · line 894
Quoted 10 lines
arm/gic.c +10
-0+/* remap the collection */ +its_send_mapc (col, true); +lpi_stats_expect (nr_cpus -1 ,8193 ); +} + +static void test_its_pending_migration (void ) +{ +struct its_device *dev; +struct its_collection *collection[2 ]; +int expected[NR_CPUS ];expected =
malloc(nr_cpus * sizeof(int));Quoted 2 lines
I know there are other places using
NR_CPUSright now that don't have to, but we shouldn't add more. Eventually I'll change the other places too.OK
Quoted 29 lines
Quoted 15 lines
arm/gic.c +14
-0+u64 pendbaser; +void *ptr; +int i; + +if (its_prerequisites (4 )) +return ; + + dev =its_create_device (2 /* dev id */ ,8 /* nb_ites */ ); +its_send_mapd (dev, true); + + collection[0 ] =its_create_collection (nr_cpus -1 , nr_cpus -1 ); + collection[1 ] =its_create_collection (nr_cpus -2 , nr_cpus -2 ); +its_send_mapc (collection[0 ], true); +its_send_mapc (collection[1 ], true);arm/gic.c static void test_its_migration(void) · line 918
Quoted 10 lines
arm/gic.c +10
-0+ +/* disable lpi at redist level */ +gicv3_lpi_rdist_ctrl (nr_cpus -1 , false); +gicv3_lpi_rdist_ctrl (nr_cpus -2 , false); + +/* even lpis are assigned to even cpu */ +for (i =0 ; i <256 ; i++) { +struct its_collection *col = i %2 ? collection[0 ] : + collection[1 ]; +int vcpu = col->target_address >>16 ;I'm lost with the even/odd (
nr_cpus- 1)/(nr_cpus - 2) stuff, and won't it swap ifnr_cpusis odd vs. even?Quoted 8 lines
Shouldn't we just have something like
pe1 = nr_cpus - 1; pe2 = nr_cpus - 2; col1 = its_create_collection(pe1, pe1); col2 = its_create_collection(pe2, pe2);without mentioning even and odd?
OK
Quoted 22 lines
Quoted 22 lines
arm/gic.c +21
-0+ +its_send_mapti (dev,8192 + i, i, col); +gicv3_lpi_set_config (8192 + i,LPI_PROP_DEFAULT ); +gicv3_lpi_set_pending_table_bit (vcpu,8192 + i, true); + } +its_send_invall (collection[0 ]); +its_send_invall (collection[1 ]); + +/* Set the PTZ bit on each pendbaser */ + + expected[nr_cpus -1 ] =128 ; + expected[nr_cpus -2 ] =128 ; + + ptr = gicv3_data.redist_base[nr_cpus -1 ] +GICR_PENDBASER ; + pendbaser =readq (ptr); +writeq (pendbaser & ~GICR_PENDBASER_PTZ , ptr); + + ptr = gicv3_data.redist_base[nr_cpus -2 ] +GICR_PENDBASER ; + pendbaser =readq (ptr); +writeq (pendbaser & ~GICR_PENDBASER_PTZ , ptr); +Quoted 8 lines
arm/gic.c static void test_its_migration(void) · line 945
Quoted 6 lines
arm/gic.c +6
-0+gicv3_lpi_rdist_ctrl (nr_cpus -1 , true); +gicv3_lpi_rdist_ctrl (nr_cpus -2 , true); + +puts ("Now migrate the VM, then press a key to continue...\n" ); + (void )getchar (); +report (true,"Migration complete" );
report_infoOK
Quoted 5 lines
arm/gic.c static void test_its_migration(void) · line 947
Quoted 3 lines
arm/gic.c +2
-0+ +mdelay (1000 );This delay needs a comment explaining why it's here.
OK
Thanks
Eric
Quoted 57 lines
Quoted 53 lines
arm/gic.c +11
-0+ + check_lpi_hits(expected); +}#endif int main (int argc,char **argv)@@ -847,6 +989,14 @@ int main(int argc, char **argv)report_prefix_push (argv[1 ]);test_its_migration ();report_prefix_pop (); + }else if (!strcmp (argv[1 ],"its-pending-migration" )) { +report_prefix_push (argv[1 ]); +test_its_pending_migration (); +report_prefix_pop (); + }else if (!strcmp (argv[1 ],"its-migrate-unmapped-collection" )) { +report_prefix_push (argv[1 ]); +test_migrate_unmapped_collection (); +report_prefix_pop (); }else if (strcmp (argv[1 ],"its-introspection" ) ==0 ) {report_prefix_push (argv[1 ]);test_its_introspection ();arm/unittests.cfg +16
-0@@ -144,6 +144,22 @@ extra_params = -machine gic-version=3 -append 'its-migration' groups = its migration arch = arm64 +[its-pending-migration] +file = gic.flat +smp = $MAX_SMP +accel = kvm +extra_params = -machine gic-version=3 -append 'its-pending-migration' +groups = its migration +arch = arm64 + +[its-migrate-unmapped-collection] +file = gic.flat +smp = $MAX_SMP +accel = kvm +extra_params = -machine gic-version=3 -append 'its-migrate-unmapped-collection' +groups = its migration +arch = arm64 + # Test PSCI emulation [psci] file = psci.flatSignature · 1 line
2.20.1Thanks,
drew
Re: [kvm-unit-tests PATCH v3 14/14] arm/arm64: ITS: pending table migration test
Hi
Quoted 55 lines · Andrew Jones
Quoted 48 lines · Auger Eric
Hi Drew,
Quoted 45 lines · Andrew Jones
Quoted 30 lines · Eric Auger
Add two new migration tests. One testing the migration of a topology where collection were unmapped. The second test checks the migration of the pending table.
Signed-off-by Eric Auger <[email protected]>---
v2 -> v3:
- tests belong to both its and migration groups--- arm/gic.c | 150 ++++++++++++++++++++++++++++++++++++++++++++++ arm/unittests.cfg | 16 +++++ 2 files changed, 166 insertions(+)arm/gic.c +1
-0@@ -195,6 +195,7 @@ static void lpi_handler(struct pt_regs *regs __unused)smp_rmb ();/* pairs with wmb in lpi_stats_expect */ lpi_stats.observed.cpu_id =smp_processor_id (); lpi_stats.observed.lpi_id = irqnr; + acked[lpi_stats.observed.cpu_id]++;smp_wmb ();/* pairs with rmb in check_lpi_stats */ }@@ -239,6 +240,18 @@ static void secondary_lpi_test(void)while (1 )wfi ();arm/gic.c static void secondary_lpi_test(void) · line 251
Quoted 10 lines
arm/gic.c +9
-0} + +static void check_lpi_hits (int *expected) +{ +int i; + +for (i =0 ; i < nr_cpus; i++) { +if (acked[i] != expected[i]) +report (false,"expected %d LPIs on PE #%d, %d observed" , + expected[i], i, acked[i]);
report_info
pass = falseQuoted 5 lines
Quoted 4 lines
arm/gic.c static void secondary_lpi_test(void) · line 253
Quoted 2 lines
arm/gic.c +2
-0+ } +report (true,"check LPI on all vcpus" );
report(pass, ...);I still don't get the issue.
Quoted 7 lines
Your messages aren't consistent and have dynamic info. You need
report_info(...); // dynamic infoif (failure_condition) { pass_boolean = false; }
report(pass_boolean, single_message_for_both_pass_and_failure);
OK understood.
Thanks!
Eric
Quoted 248 lines
Quoted 248 lines
Quoted 5 lines
Quoted 5 lines
arm/gic.c +1
-0+}#endif Quoted 10 lines
arm/gic.c static void gic_test_mmio(void) · line 611
Quoted 7 lines
arm/gic.c +2
-0static void gicv2_ipi_send_self (void )@@ -594,6 +607,8 @@ static void gic_test_mmio(void)static void test_its_introspection (void ) {}static void test_its_trigger (void ) {}static void test_its_migration (void ) {} +static void test_migrate_unmapped_collection (void ) {} +static void test_its_pending_migration (void ) {}I'm not sure what's worse. This pile of stubs or one #ifdef in
main()wrapping all the calls.Those stubs now are in the arm header.
Quoted 8 lines
Quoted 8 lines
arm/gic.c +0
-0#else /* __arch64__ */ @@ -666,6 +681,18 @@ static bool its_prerequisites(int nb_cpus)return false; }Quoted 8 lines
arm/gic.c static bool its_prerequisites(int nb_cpus) · line 687
Quoted 4 lines
arm/gic.c +4
-0+static void set_lpi (struct its_device *dev,u32 eventid,u32 physid, +struct its_collection *col) +{ +if (!dev || !col)I don't think col can be null, and this doesn't look like the right place to check if dev is null. If we're bothiner to call
set_lpi, then I think we should already expect dev to be good to go.put an
assert()insteadQuoted 30 lines
Quoted 30 lines
arm/gic.c +22
-0+report_abort ("wrong device or collection" ); + +its_send_mapti (dev, physid, eventid, col); + +gicv3_lpi_set_config (physid,LPI_PROP_DEFAULT ); +its_send_invall (col); +} +/* * Setup the configuration for those mappings: * dev_id=2 event=20 -> vcpu 3, intid=8195 @@ -806,6 +833,121 @@ static void test_its_migration(void)its_send_int (dev7,255 );check_lpi_stats (); } + +static void test_migrate_unmapped_collection (void ) +{ +struct its_collection *col; +struct its_device *dev2, *dev7; +u8 config; + +if (its_setup1 ()) +return ; + + col =its_create_collection (nr_cpus -1 , nr_cpus -1 ); + dev2 =its_get_device (2 ); + dev7 =its_get_device (7 ); +Quoted 8 lines
arm/gic.c static void test_its_migration(void) · line 855
Quoted 6 lines
arm/gic.c +6
-0+/* MAPTI with the collection unmapped */ +set_lpi (dev2,0 ,8192 , col); + +puts ("Now migrate the VM, then press a key to continue...\n" ); + (void )getchar (); +report (true,"Migration complete" );
report_infoyep
Quoted 43 lines
Quoted 30 lines
arm/gic.c +29
-0+ +/* on the destination, map the collection */ +its_send_mapc (col, true); + +lpi_stats_expect (2 ,8196 ); +its_send_int (dev7,255 ); +check_lpi_stats (); + + config =gicv3_lpi_get_config (8192 ); +report (config ==LPI_PROP_DEFAULT , +"Config of LPI 8192 was properly migrated" ); + +lpi_stats_expect (nr_cpus -1 ,8192 ); +its_send_int (dev2,0 ); +check_lpi_stats (); + +/* unmap the collection */ +its_send_mapc (col, false); + +lpi_stats_expect (-1 ,-1 ); +its_send_int (dev2,0 ); +check_lpi_stats (); + +/* remap event 0 onto lpiid 8193 */ +set_lpi (dev2,0 ,8193 , col); +lpi_stats_expect (-1 ,-1 ); +its_send_int (dev2,0 ); +check_lpi_stats (); +arm/gic.c static void test_its_migration(void) · line 894
Quoted 10 lines
arm/gic.c +10
-0+/* remap the collection */ +its_send_mapc (col, true); +lpi_stats_expect (nr_cpus -1 ,8193 ); +} + +static void test_its_pending_migration (void ) +{ +struct its_device *dev; +struct its_collection *collection[2 ]; +int expected[NR_CPUS ];expected =
malloc(nr_cpus * sizeof(int));Quoted 2 lines
I know there are other places using
NR_CPUSright now that don't have to, but we shouldn't add more. Eventually I'll change the other places too.OK
Quoted 29 lines
Quoted 15 lines
arm/gic.c +14
-0+u64 pendbaser; +void *ptr; +int i; + +if (its_prerequisites (4 )) +return ; + + dev =its_create_device (2 /* dev id */ ,8 /* nb_ites */ ); +its_send_mapd (dev, true); + + collection[0 ] =its_create_collection (nr_cpus -1 , nr_cpus -1 ); + collection[1 ] =its_create_collection (nr_cpus -2 , nr_cpus -2 ); +its_send_mapc (collection[0 ], true); +its_send_mapc (collection[1 ], true);arm/gic.c static void test_its_migration(void) · line 918
Quoted 10 lines
arm/gic.c +10
-0+ +/* disable lpi at redist level */ +gicv3_lpi_rdist_ctrl (nr_cpus -1 , false); +gicv3_lpi_rdist_ctrl (nr_cpus -2 , false); + +/* even lpis are assigned to even cpu */ +for (i =0 ; i <256 ; i++) { +struct its_collection *col = i %2 ? collection[0 ] : + collection[1 ]; +int vcpu = col->target_address >>16 ;I'm lost with the even/odd (
nr_cpus- 1)/(nr_cpus - 2) stuff, and won't it swap ifnr_cpusis odd vs. even?Quoted 8 lines
Shouldn't we just have something like
pe1 = nr_cpus - 1; pe2 = nr_cpus - 2; col1 = its_create_collection(pe1, pe1); col2 = its_create_collection(pe2, pe2);without mentioning even and odd?
OK
Quoted 22 lines
Quoted 22 lines
arm/gic.c +21
-0+ +its_send_mapti (dev,8192 + i, i, col); +gicv3_lpi_set_config (8192 + i,LPI_PROP_DEFAULT ); +gicv3_lpi_set_pending_table_bit (vcpu,8192 + i, true); + } +its_send_invall (collection[0 ]); +its_send_invall (collection[1 ]); + +/* Set the PTZ bit on each pendbaser */ + + expected[nr_cpus -1 ] =128 ; + expected[nr_cpus -2 ] =128 ; + + ptr = gicv3_data.redist_base[nr_cpus -1 ] +GICR_PENDBASER ; + pendbaser =readq (ptr); +writeq (pendbaser & ~GICR_PENDBASER_PTZ , ptr); + + ptr = gicv3_data.redist_base[nr_cpus -2 ] +GICR_PENDBASER ; + pendbaser =readq (ptr); +writeq (pendbaser & ~GICR_PENDBASER_PTZ , ptr); +Quoted 8 lines
arm/gic.c static void test_its_migration(void) · line 945
Quoted 6 lines
arm/gic.c +6
-0+gicv3_lpi_rdist_ctrl (nr_cpus -1 , true); +gicv3_lpi_rdist_ctrl (nr_cpus -2 , true); + +puts ("Now migrate the VM, then press a key to continue...\n" ); + (void )getchar (); +report (true,"Migration complete" );
report_infoOK
Quoted 5 lines
arm/gic.c static void test_its_migration(void) · line 947
Quoted 3 lines
arm/gic.c +2
-0+ +mdelay (1000 );This delay needs a comment explaining why it's here.
OK
Thanks
Eric
Quoted 51 lines
Quoted 51 lines
arm/gic.c +11
-0+ + check_lpi_hits(expected); +}#endif int main (int argc,char **argv)@@ -847,6 +989,14 @@ int main(int argc, char **argv)report_prefix_push (argv[1 ]);test_its_migration ();report_prefix_pop (); + }else if (!strcmp (argv[1 ],"its-pending-migration" )) { +report_prefix_push (argv[1 ]); +test_its_pending_migration (); +report_prefix_pop (); + }else if (!strcmp (argv[1 ],"its-migrate-unmapped-collection" )) { +report_prefix_push (argv[1 ]); +test_migrate_unmapped_collection (); +report_prefix_pop (); }else if (strcmp (argv[1 ],"its-introspection" ) ==0 ) {report_prefix_push (argv[1 ]);test_its_introspection ();arm/unittests.cfg +16
-0@@ -144,6 +144,22 @@ extra_params = -machine gic-version=3 -append 'its-migration' groups = its migration arch = arm64 +[its-pending-migration] +file = gic.flat +smp = $MAX_SMP +accel = kvm +extra_params = -machine gic-version=3 -append 'its-pending-migration' +groups = its migration +arch = arm64 + +[its-migrate-unmapped-collection] +file = gic.flat +smp = $MAX_SMP +accel = kvm +extra_params = -machine gic-version=3 -append 'its-migrate-unmapped-collection' +groups = its migration +arch = arm64 + # Test PSCI emulation [psci] file = psci.flatSignature · 1 line
2.20.1
Quoted 8 lines
Quoted 8 lines
Quoted 6 lines
Quoted 2 lines
Thanks,
drew
_______________________________________________
kvmarm mailing list
[email protected]
https://lists.cs.columbia.edu/mailman/listinfo/kvmarm