Re: [External] Re: [PATCH] hw/block/nvme: add smart_critical_warning property

[zhenwei pi]

On 1/11/21 5:21 PM, Klaus Jensen wrote:
> On Jan 11 10:14, Philippe Mathieu-Daudé wrote:
> > On 1/11/21 8:50 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.
> > >
> > > Test with this patch:
> > > 1, append 'smart_critical_warning=16' for nvme parameters.
> > > 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 | 4 ++++
> > >   hw/block/nvme.h | 1 +
> > >   2 files changed, 5 insertions(+)
> > >
> > > diff --git a/hw/block/nvme.c b/hw/block/nvme.c
> > > index 27d2c72716..2f0bcac91c 100644
> > > --- a/hw/block/nvme.c
> > > +++ b/hw/block/nvme.c
> > > @@ -1215,6 +1215,8 @@ 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->params.smart_critical_warning;
> > > +
> > >       smart.data_units_read[0] = cpu_to_le64(DIV_ROUND_UP(stats.units_read,
> > >                                                           1000));
> > >       smart.data_units_written[0] = 
> > > cpu_to_le64(DIV_ROUND_UP(stats.units_written,
> > > @@ -2824,6 +2826,8 @@ static Property nvme_props[] = {
> > >       DEFINE_PROP_UINT32("aer_max_queued", NvmeCtrl, params.aer_max_queued, 64),
> > >       DEFINE_PROP_UINT8("mdts", NvmeCtrl, params.mdts, 7),
> > >       DEFINE_PROP_BOOL("use-intel-id", NvmeCtrl, params.use_intel_id, false),
> > > +    DEFINE_PROP_UINT8("smart_critical_warning", NvmeCtrl,
> > > +                      params.smart_critical_warning, 0),
> > >       DEFINE_PROP_END_OF_LIST(),
> > >   };
> > >   
> > > diff --git a/hw/block/nvme.h b/hw/block/nvme.h
> > > index e080a2318a..76684f5ac0 100644
> > > --- a/hw/block/nvme.h
> > > +++ b/hw/block/nvme.h
> > > @@ -16,6 +16,7 @@ typedef struct NvmeParams {
> > >       uint32_t aer_max_queued;
> > >       uint8_t  mdts;
> > >       bool     use_intel_id;
> > > +    uint8_t  smart_critical_warning;
> > >   } NvmeParams;
> > >   
> > >   typedef struct NvmeAsyncEvent {
> >
> > This is an easy way to achieve your goal.
> >
> > However a better way is to add a QMP command to
> > change NvmeCtrl->temperature.
> >
> > See for example tmp105_initfn() in hw/misc/tmp105.c
> > and qmp_tmp105_set_temperature() in tests/qtest/tmp105-test.c.
>
> Nice.
>
> +1 for this approach.

Using QMP command to change NvmeCtrl->temperature only triggers 
NVME_SMART_TEMPERATURE warning, it's OK to test the work flow of uplayer 
software, but it's not enough to test all the cases of each warning.

From NVMe version 1.3 to 1.4, a new bit definition has been added(bit 
5, Persistent Memory Region has become read-only or unreliable). Before 
we really hit this warning on a physical disk, we can use QEMU to test 
this feature(maybe another new feature in the future).

I don't disagree "add a QMP command" solution, but I think QEMU should 
be able to emulate all of the warnings(not only temperature).

--
zhenwei pi