| Message ID | 20210114072251.334304-4-pizhenwei@bytedance.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | support NVMe smart critial warning injection | expand |
On Jan 14 15:22, zhenwei pi wrote: > There is a very low probability that hitting physical NVMe disk > hardware critical warning case, it's hard to write & test a monitor > agent service. > > For debugging purposes, add a new 'smart_critical_warning' property > to emulate this situation. > > The orignal version of this change is implemented by adding a fixed > property which could be initialized by QEMU command line. Suggested > by Philippe & Klaus, rework like current version. > > Test with this patch: > 1, change smart_critical_warning property for a running VM: > #virsh qemu-monitor-command nvme-upstream '{ "execute": "qom-set", > "arguments": { "path": "/machine/peripheral-anon/device[0]", > "property": "smart_critical_warning", "value":16 } }' > 2, run smartctl in guest > #smartctl -H -l error /dev/nvme0n1 > > === START OF SMART DATA SECTION === > SMART overall-health self-assessment test result: FAILED! > - volatile memory backup device has failed > > Signed-off-by: zhenwei pi <pizhenwei@bytedance.com> > --- > hw/block/nvme.c | 40 ++++++++++++++++++++++++++++++++++++++++ > hw/block/nvme.h | 1 + > 2 files changed, 41 insertions(+) > > diff --git a/hw/block/nvme.c b/hw/block/nvme.c > index f361103bb4..ce9a9c9023 100644 > --- a/hw/block/nvme.c > +++ b/hw/block/nvme.c > @@ -1214,6 +1214,7 @@ static uint16_t nvme_smart_info(NvmeCtrl *n, uint8_t rae, uint32_t buf_len, > } > > trans_len = MIN(sizeof(smart) - off, buf_len); > + smart.critical_warning = n->smart_critical_warning; > > smart.data_units_read[0] = cpu_to_le64(DIV_ROUND_UP(stats.units_read, > 1000)); > @@ -2826,6 +2827,41 @@ static Property nvme_props[] = { > DEFINE_PROP_END_OF_LIST(), > }; > > + > +static void nvme_get_smart_warning(Object *obj, Visitor *v, const char *name, > + void *opaque, Error **errp) > +{ > + NvmeCtrl *s = NVME(obj); > + uint8_t value = s->smart_critical_warning; > + > + visit_type_uint8(v, name, &value, errp); > +} > + > +static void nvme_set_smart_warning(Object *obj, Visitor *v, const char *name, > + void *opaque, Error **errp) > +{ > + NvmeCtrl *s = NVME(obj); > + uint8_t value, cap = 0; > + uint64_t pmr_cap = CAP_PMR_MASK; > + > + if (!visit_type_uint8(v, name, &value, errp)) { > + return; > + } > + > + cap = NVME_SMART_SPARE | NVME_SMART_TEMPERATURE | NVME_SMART_RELIABILITY > + | NVME_SMART_MEDIA_READ_ONLY | NVME_SMART_FAILED_VOLATILE_MEDIA; > + if (s->bar.cap & (pmr_cap << CAP_PMR_SHIFT)) { > + cap |= NVME_SMART_PMR_UNRELIABLE; > + } Looks like an NVME_CAP_PMRS(cap) macro is missing in include/block/nvme.h. I have added it in another PMR series under review, but you can add it here as well instead of manually doing the shift and check. > + > + if ((value & cap) != value) { > + error_setg(errp, "unsupported smart critical warning value"); > + return; > + } > + > + s->smart_critical_warning = value; > +} > + > static const VMStateDescription nvme_vmstate = { > .name = "nvme", > .unmigratable = 1, > @@ -2856,6 +2892,10 @@ static void nvme_instance_init(Object *obj) > "bootindex", "/namespace@1,0", > DEVICE(obj)); > } > + > + object_property_add(obj, "smart_critical_warning", "uint8", > + nvme_get_smart_warning, > + nvme_set_smart_warning, NULL, NULL); > } > > static const TypeInfo nvme_info = { > diff --git a/hw/block/nvme.h b/hw/block/nvme.h > index e080a2318a..64e3497244 100644 > --- a/hw/block/nvme.h > +++ b/hw/block/nvme.h > @@ -139,6 +139,7 @@ typedef struct NvmeCtrl { > uint64_t timestamp_set_qemu_clock_ms; /* QEMU clock time */ > uint64_t starttime_ms; > uint16_t temperature; > + uint8_t smart_critical_warning; > > HostMemoryBackend *pmrdev; > > -- > 2.25.1 > >
On 1/14/21 8:22 AM, zhenwei pi wrote: > There is a very low probability that hitting physical NVMe disk > hardware critical warning case, it's hard to write & test a monitor > agent service. > > For debugging purposes, add a new 'smart_critical_warning' property > to emulate this situation. > > The orignal version of this change is implemented by adding a fixed > property which could be initialized by QEMU command line. Suggested > by Philippe & Klaus, rework like current version. > > Test with this patch: > 1, change smart_critical_warning property for a running VM: > #virsh qemu-monitor-command nvme-upstream '{ "execute": "qom-set", > "arguments": { "path": "/machine/peripheral-anon/device[0]", > "property": "smart_critical_warning", "value":16 } }' > 2, run smartctl in guest > #smartctl -H -l error /dev/nvme0n1 > > === START OF SMART DATA SECTION === > SMART overall-health self-assessment test result: FAILED! > - volatile memory backup device has failed > > Signed-off-by: zhenwei pi <pizhenwei@bytedance.com> > --- > hw/block/nvme.c | 40 ++++++++++++++++++++++++++++++++++++++++ > hw/block/nvme.h | 1 + > 2 files changed, 41 insertions(+) ... > +static void nvme_set_smart_warning(Object *obj, Visitor *v, const char *name, > + void *opaque, Error **errp) > +{ > + NvmeCtrl *s = NVME(obj); > + uint8_t value, cap = 0; > + uint64_t pmr_cap = CAP_PMR_MASK; > + > + if (!visit_type_uint8(v, name, &value, errp)) { > + return; > + } > + > + cap = NVME_SMART_SPARE | NVME_SMART_TEMPERATURE | NVME_SMART_RELIABILITY > + | NVME_SMART_MEDIA_READ_ONLY | NVME_SMART_FAILED_VOLATILE_MEDIA; > + if (s->bar.cap & (pmr_cap << CAP_PMR_SHIFT)) { > + cap |= NVME_SMART_PMR_UNRELIABLE; > + } > + > + if ((value & cap) != value) { > + error_setg(errp, "unsupported smart critical warning value"); More useful: error_setg(errp, "unsupported smart critical warning bits: 0x%x", value & ~cap); Regardless: Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com> Thanks!
On Thu, Jan 14, 2021 at 03:22:50PM +0800, zhenwei pi wrote: > +static void nvme_get_smart_warning(Object *obj, Visitor *v, const char *name, > + void *opaque, Error **errp) > +{ > + NvmeCtrl *s = NVME(obj); With only one exception, all variables of type 'NvmeCtrl' in this program are called 'n', so let's keep that consistency please. Otherwise, this looks fine. > + uint8_t value = s->smart_critical_warning; > + > + visit_type_uint8(v, name, &value, errp); > +} > + > +static void nvme_set_smart_warning(Object *obj, Visitor *v, const char *name, > + void *opaque, Error **errp) > +{ > + NvmeCtrl *s = NVME(obj);
diff --git a/hw/block/nvme.c b/hw/block/nvme.c index f361103bb4..ce9a9c9023 100644 --- a/hw/block/nvme.c +++ b/hw/block/nvme.c @@ -1214,6 +1214,7 @@ static uint16_t nvme_smart_info(NvmeCtrl *n, uint8_t rae, uint32_t buf_len, } trans_len = MIN(sizeof(smart) - off, buf_len); + smart.critical_warning = n->smart_critical_warning; smart.data_units_read[0] = cpu_to_le64(DIV_ROUND_UP(stats.units_read, 1000)); @@ -2826,6 +2827,41 @@ static Property nvme_props[] = { DEFINE_PROP_END_OF_LIST(), }; + +static void nvme_get_smart_warning(Object *obj, Visitor *v, const char *name, + void *opaque, Error **errp) +{ + NvmeCtrl *s = NVME(obj); + uint8_t value = s->smart_critical_warning; + + visit_type_uint8(v, name, &value, errp); +} + +static void nvme_set_smart_warning(Object *obj, Visitor *v, const char *name, + void *opaque, Error **errp) +{ + NvmeCtrl *s = NVME(obj); + uint8_t value, cap = 0; + uint64_t pmr_cap = CAP_PMR_MASK; + + if (!visit_type_uint8(v, name, &value, errp)) { + return; + } + + cap = NVME_SMART_SPARE | NVME_SMART_TEMPERATURE | NVME_SMART_RELIABILITY + | NVME_SMART_MEDIA_READ_ONLY | NVME_SMART_FAILED_VOLATILE_MEDIA; + if (s->bar.cap & (pmr_cap << CAP_PMR_SHIFT)) { + cap |= NVME_SMART_PMR_UNRELIABLE; + } + + if ((value & cap) != value) { + error_setg(errp, "unsupported smart critical warning value"); + return; + } + + s->smart_critical_warning = value; +} + static const VMStateDescription nvme_vmstate = { .name = "nvme", .unmigratable = 1, @@ -2856,6 +2892,10 @@ static void nvme_instance_init(Object *obj) "bootindex", "/namespace@1,0", DEVICE(obj)); } + + object_property_add(obj, "smart_critical_warning", "uint8", + nvme_get_smart_warning, + nvme_set_smart_warning, NULL, NULL); } static const TypeInfo nvme_info = { diff --git a/hw/block/nvme.h b/hw/block/nvme.h index e080a2318a..64e3497244 100644 --- a/hw/block/nvme.h +++ b/hw/block/nvme.h @@ -139,6 +139,7 @@ typedef struct NvmeCtrl { uint64_t timestamp_set_qemu_clock_ms; /* QEMU clock time */ uint64_t starttime_ms; uint16_t temperature; + uint8_t smart_critical_warning; HostMemoryBackend *pmrdev;
There is a very low probability that hitting physical NVMe disk hardware critical warning case, it's hard to write & test a monitor agent service. For debugging purposes, add a new 'smart_critical_warning' property to emulate this situation. The orignal version of this change is implemented by adding a fixed property which could be initialized by QEMU command line. Suggested by Philippe & Klaus, rework like current version. Test with this patch: 1, change smart_critical_warning property for a running VM: #virsh qemu-monitor-command nvme-upstream '{ "execute": "qom-set", "arguments": { "path": "/machine/peripheral-anon/device[0]", "property": "smart_critical_warning", "value":16 } }' 2, run smartctl in guest #smartctl -H -l error /dev/nvme0n1 === START OF SMART DATA SECTION === SMART overall-health self-assessment test result: FAILED! - volatile memory backup device has failed Signed-off-by: zhenwei pi <pizhenwei@bytedance.com> --- hw/block/nvme.c | 40 ++++++++++++++++++++++++++++++++++++++++ hw/block/nvme.h | 1 + 2 files changed, 41 insertions(+)