On Thu, 17 Nov 2022 15:42:43 -0800 Alison Schofield <[email protected]> wrote:
> On Wed, Nov 16, 2022 at 01:03:45PM +0000, Jonathan Cameron wrote: > > On Thu, 10 Nov 2022 19:20:07 -0800 > > [email protected] wrote: > > > > > From: Alison Schofield <[email protected]> > > > > > > The --media-errors option to 'cxl list' retrieves poison lists > > > from memory devices (supporting the capability) and displays > > > the returned media-error records in the cxl list json. This > > > option can apply to memdevs or regions. > > > > > > Signed-off-by: Alison Schofield <[email protected]> > > > --- > > > Documentation/cxl/cxl-list.txt | 64 ++++++++++++++++++++++++++++++++++ > > > cxl/filter.c | 2 ++ > > > cxl/filter.h | 1 + > > > cxl/list.c | 2 ++ > > > 4 files changed, 69 insertions(+) > > > > > > diff --git a/Documentation/cxl/cxl-list.txt > > > b/Documentation/cxl/cxl-list.txt > > > index 14a2b4bb5c2a..24a0cf97cef2 100644 > > > --- a/Documentation/cxl/cxl-list.txt > > > +++ b/Documentation/cxl/cxl-list.txt > > > @@ -344,6 +344,70 @@ OPTIONS > > > --region:: > > > Specify CXL region device name(s), or device id(s), to filter the > > > listing. > > > > > > +-a:: > > > +--media-errors:: > > > + Include media-error information. The poison list is retrieved > > > + from the device(s) and media error records are added to the > > > + listing. This option applies to memdevs and regions where > > > + devices support the poison list capability. > > > > I'm not sure media errors is a good name. The poison doesn't have to > > originate > > in the device. Given we are logging poison with "external" as the source > > those definitely don't come from the device and may have nothing to do > > with 'media' as such. > > > > Why not just call it poison? > > > --media-errors probably originated from ndctl tool which used > that same option name, but it fits in with the CXL Spec language. > > The CXL Spec calls the records returned from the 'Get Poison List' > command Media Error Records. It refers to poison as media errors. > So, here, in a command that lists things - the thing(s) being listed > is(are) 'media error record(s)'. > > I see what you're saying about 'External' source. Does that mean > an 'External' source caused an actual media error? Hmm. I suspect this all evolved. An External source need not have anything to do with media (could be corruption in some random cache or on interconnect or even that a link collapsed potentially). Ah well, I'm fine with any naming you prefer. No idea if the NVDIMM equivalent has a the same issue with externally generated poison. > > So, that 'Why not poison?' answer. I'm easily swayed either way. > Would you suggest: > > > + > > > +---- > > > +# cxl list -m mem11 --media-errors > > cxl list -m mem1 --poison > > > > + "media_errors":{ > > > + "nr_media_errors":1, > > > + "media_error_records":[ > > and rename the fields above: > "poison_errors" > "nr_poison_errors" > "poison_error_records" > > That works for me, but if it's going to confuse people familiar with other similar cases, then I don't mind the original naming that much. Jonathan
