Hi, On 8/27/22 15:49, Hans de Goede wrote: > hI, > > On 8/27/22 13:23, Arvid Norlander wrote: >> Hi, >> >> On 2022-08-26 13:46, Hans de Goede wrote: >>> Hi All, >>> >>> [...] >>> >>> Arvid, I wonder if instead of using disable_backlight_sysfs_if >>> you can try: >>> >>> 0. Remove disable_backlight_sysfs_if from cmdline / quirk >>> 1. Adding acpi_backlight=native to the kernel commandline >>> 2. In toshiba_acpi_resume() add a HCI_PANEL_POWER_ON PANEL_ON >>> >>> and see if that also fixes things ? >>> >> Yes, this works. > > Great, thank you for testing this! > > In hindsight the disable_backlight_sysfs_if flag was a mistake > and I should have fixed this differently (I wrote the code adding > that flag). And now it is sorta getting in the way of cleaning > up the backlight handling. So IMHO removing disable_backlight_sysfs_if > is the best thing to do here. > >> I do not have a patch for this (I assume it >> would involve creating quirk tables, checking for support for >> HCI_PANEL_POWER_ON, etc). I simply hard coded the call in for >> the test. I very much doubt I will have time to code this in >> the near future as well. > > No problem I will prepare a patch series for you to test. Note > this will be on top of my other backlight cleanups, so I > will just send you an email pointing to a git branch to test, > I hope this will be ok? Arvid, here is a git branch with my backlight-refactor for you to test: https://github.com/jwrdegoede/linux-sunxi/commits/backlight-refactor-for-arvid If you can give this a test spin (without any special kernel commandline options) then that would be great. >> However, do we know what the other Toshiba's that need this >> quirk also supports HCI_PANEL_POWER_ON? I obviously can only >> test the Z830 that I own. > > It seems that all models which need this are all from the same > generation so I would expect the same fix to work. If I get > regression reports from users after my cleanup series lands > I can then take a closer look at the DSDT tables of the > other models if necessary. Quick update on this I've taken a look at the DSTD's _BCM method which is the magic call done by acpi_video on resume which turns the panel back on and in both a R700 as well as a R830 it ends up doing this: SMBR (0xFF00, 0x2A, Local0, 0x00, 0xB2) Which translates to: HCI_SET HCI_BRIGHTNESS <brightness-level> Note that in this case brightness-level used by the acpi_video code is the initial brightness at power-on since acpi_video is not used to actually control the brightness. Which makes using HCI_PANEL_POWER_ON a better idea since that does not touches the brightness. So I believe that replacing disable_backlight_sysfs_if with a HCI call by toshiba_acpi on resume should work on all affected models. Regards, Hans
