On Sat, Mar 2, 2024 at 1:38 PM Hans de Goede <hdegoede@xxxxxxxxxx> wrote: > On 2/29/24 21:57, Pali Rohár wrote: ... > But the only real problem you have pointed out is that > if people *manually* rmmod i2c-i801 that then after *manually* > modprobing i2c-i801 again the i2c_client for the lis3lv02d chip > is not automatically re-instated, instead they will need to > also manually reload the dell-smo8800 driver. Which IMHO > really is not an issue since they are already manually messing > with drivers anyways. Actually we probably can add some code to the i2c core to list the clients that were instantiated in a way that can't be done again after removing the respective bus driver. But this is just an idea on how to inform users what exactly can go wrong after the rmmod/modprobe cycle, I haven't investigated this anyhow. > And note that even that problem could be fixed by using > bus-notifiers as Andy suggested. IMHO using notifiers here is > overkill. But if you are ok with moving this code out of i2c-i801 > and intel dell-smo8800 if I use notifiers in the next version so > that things will keep working even after a *manual* rmmod of > i2c-i801 then I'll do so for v3 of the patchset. -- With Best Regards, Andy Shevchenko
