On Sun, Mar 24, 2024 at 04:39:53PM -0700, Christoph Hellwig wrote: > On Fri, Mar 22, 2024 at 01:09:51PM -0400, Brian Foster wrote: > > Ok, I guess the reason I was only seeing this in one particular VM is > > that for whatever reason, something happens to invoke pktsetup during > > boot on that guest. If I boot up a separate VM/distro and run 'pktsetup > > 0 /dev/sr0,' I see the same splat.. > > pktcdvd is completely unmaintained and in horrible shape unfortunately, > and no one is interested in caring for it. > > I can only recommend to not build it into any kernel you can care about. > Heh, Ok thanks. I was poking around the code a bit just to try and figure if there was some easy/incremental improvement to be made here, but if so, I'm not familiar enough with the code to see it. I was particularly wondering why the multi-open occurs across the setup and pktcdvd device open and if that can be avoided. It looks like the main difference is the nonblock flag. From digging down into cdrom_open(), that is what controls whether or not the drive is opened for data or control purposes (i.e., so dictates whether media is read or not), and so that seems to explain why a blocking open would be deferred to pktcdvd device open. Anyways, it does seem like the warning itself is benign because the pktcdvd and backing device will be separate disks. Brian
