On 11/6/26 5:38 PM, Danilo Krummrich wrote: > On Sat Nov 7, 2026 at 1:39 AM CET, John Hubbard wrote: >> Specifically, (as you mention above), I had to add "retained/" to the >> directory path in order to retrieve the logs. That is awkward, because >> that means that if I ask for gsp_keep_logs=1, now I have to know/remember >> that they are in: >> >> /sys/kernel/debug/nova-core/retained/<BDF>/ >> >> instead of: >> >> /sys/kernel/debug/nova-core/<BDF> >> >> We should try hard to avoid such complications, especially given that >> debugging is often done when one has a thousand other complicated things >> going on at the same time, late at night, too. > > It doesn't bother me too much, but I think the concern is fair. > > The retained directory keeps a clear separation between live buffers and > retained buffers, but I don't think that's a strong reason to keep it. > > In nouveau we don't have that separation either, but nouveau also keeps the > retained buffers until module unload. I.e. after the first retention nouveau > never exposes new buffers (including the live ones) unless you reload the > module. This works, but can be rather unexpected when e.g. doing multiple > rebinds via sysfs. > > Rebinding via sysfs can be especially useful in multi-GPU setups, where you > don't want to unbind all GPUs by unloading the module. > > So, getting rid of retained/ is fine, but let's not follow what nouveau does. > Instead, let's make the same files point to the live buffers when the device > is > bound and otherwise let them point at the retained buffers.
This sounds right and proper, yes. :) thanks, -- John Hubbard
