[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <9e91ba36-0a4f-4613-e45c-7919205d8704@grimberg.me>
Date: Tue, 11 Jul 2023 17:39:51 +0300
From: Sagi Grimberg <sagi@...mberg.me>
To: Greg KH <gregkh@...uxfoundation.org>
Cc: Christoph Hellwig <hch@....de>,
Linux regressions mailing list <regressions@...ts.linux.dev>,
Pankaj Raghav <p.raghav@...sung.com>,
Keith Busch <kbusch@...nel.org>,
Bagas Sanjaya <bagasdotme@...il.com>,
Jens Axboe <axboe@...nel.dk>,
"Clemens S." <cspringsguth@...il.com>,
Martin Belanger <martin.belanger@...l.com>,
Chaitanya Kulkarni <kch@...dia.com>,
John Meneghini <jmeneghi@...hat.com>,
Hannes Reinecke <hare@...e.de>,
Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
Linux NVMe <linux-nvme@...ts.infradead.org>,
Kanchan Joshi <joshi.k@...sung.com>,
Javier Gonzalez <javier.gonz@...sung.com>,
박진환 <jh.i.park@...sung.com>,
Linus Torvalds <torvalds@...ux-foundation.org>
Subject: Re: Fwd: Need NVME QUIRK BOGUS for SAMSUNG MZ1WV480HCGL-000MV
(Samsung SM-953 Datacenter SSD)
>>>> Well, that "They keep pumping out more and more devices with the same
>>>> breakage" and the "new device" comment from Pankaj below bear the
>>>> question: should we stop trying to play "whack a mole" with all those
>>>> quirk entries and handle devices with duplicate ids just like Windows does?
>>>
>>> As far as I can tell Windows completely ignores the IDs. Which, looking
>>> back, I'd love to be able to do as well, but they are already used
>>> by udev for the /dev/disk/by-id/ links. Those are usually not used
>>> on desktop systems, as they use the file system labels and UUIDs, but
>>> that doesn't work for non-file system uses.
>>>
>>> And all this has been working really well with the good old enterprise
>>> SSDs, it's just that the cheap consumer devices keep fucking it up.
>>>
>>> If we'd take it away now we'd break existing users, which puts us between
>>> a rock and a hard place.
>>
>> Maybe the compromise would be to add a modparam that tells the driver
>> to ignore it altogether (like allow_bogus_identifiers) that would
>> default to false. Then people can just workaround the problem instead
>> of having the back-and-fourth with the vendor?
>>
>
> Module parameters do not work on a per-device basis, sorry. This isn't
> the 1990's anymore, please do not attempt to add new ones :)
Don't get me wrong, I don't like adding this. But the source of this
is that there are simply too many breakages of non-compliant consumer
drives out there that maybe a compromise would be "globally relax
compliance check in this specific area" as a workaround.
Right now each time this issue is seen in the wild, the only
resolution is either the vendor fixing it, or the driver adds
a quirk, which is positive and exactly what we want. But more
and more users complain, and there is no immediate workaround.
Powered by blists - more mailing lists