ChanServ changed the topic of #aarch64-laptops to: Linux support for AArch64 Laptops (Asus NovaGo TP370QL - HP Envy x2 - Lenovo Mixx 630 - Lenovo Yoga C630)
<jenneron[m]>
also, i still couldn't get modem to work
<jenneron[m]>
has anyone actually tried it?
<jenneron[m]>
on flex 5g
<jenneron[m]>
and i sometimes get [drm:dpu_encoder_frame_done_timeout:2388] [dpu error]enc31 frame done timeout
<steev>
yeah, that's not uncommon, i've seen that before, used to see it a lot on the c630, but i don't recall what abnihav did to fix it, or if we ever did or if it just became so rare that i ignored it
<steev>
i haven't tried modem at all, i'm not rich enough to have a mobile connection on my laptop
<jenneron[m]>
i'm daily driving this laptop and this thing is a little annoying
<jenneron[m]>
steev: i'm lucky that it's pretty cheap in my country
<jenneron[m]>
i pay 6$ for 1 TB / month
<jenneron[m]>
however this tariff is not for everyone
<qzed>
jhovold: from what I can tell from the locks, there's a problem initializing qcom_scm
<qzed>
qcom_scm: firmware:scm: error -ENXIO: IRQ index 0 not found
<qzed>
so my guess is that the custom efivar ops aren't even registered
<qzed>
that would fit __efivars being NULL
<jhovold>
yeah, that's what it looks like, but the series I linked to above would have added a check for a NULL __efivars at mount time which should have prevented the NULL deref
<qzed>
right, I know too little about fs stuff, so I assume that an error in efivarfs_fill_super() ensures that efivarfs_kill_sb() is never called?
<jhovold>
I would have thought so, but perhaps steev is using my series and that's where the bug really is
<qzed>
I'm just wondering how this works, because the efivarfs_kill_sb() seems like a cleanup function? so I guess something in the setup errors out (which could be due to your check)
<qzed>
so maybe a similar check is required in efivarfs_kill_sb()?
<jhovold>
yeah, that's what I fear. Checking now
<jhovold>
qzed, ardb: yes, that's it. My patch was buggy, preparing a fix
<jhovold>
steev: this should address the efivarfs NULL-deref (but not the scm driver error):
<juergh>
abelvesa, I'm getting 'UBSAN: shift-out-of-bounds in drivers/soc/qcom/llcc-qcom.c:772:45' on an x13s. in my setup slice_id goes all the way up to 36.
<abelvesa>
juergh: which kernel version are you running with ?
<abelvesa>
juergh: wait, it is in the pastebin
<juergh>
abelvesa, linux-next
<juergh>
well that is an ubuntu abomination of 6.2-rc4 with some next patches
iivanov has quit [Ping timeout: 480 seconds]
<steev>
jhovold: qzed: will give that a spin - as for how... well, i have no idea what i'm doing, so i had added the qsee-com entry but not an scm entry because i don't know what scm should be, and previous iteration didn't require it, but that gave the same null deref, so i simply backed out the dts changes, but left qseecom and efivars enabled (but nothing in the device tree, afaik)
<steev>
and yeah, it's with the forthcoming v2 series
<jhovold>
changes since last branch (rc1), include:
<jhovold>
- rtc support
<jhovold>
- display support updated (v7)
<jhovold>
- pmic glink updated (v2)
<jhovold>
- johan_defconfig updated
<jhovold>
- sound enabled in dts, but not in config due to incomplete driver support
<jhovold>
- mdss iommu identity domain
<jhovold>
- efivarfs fixes
<steev>
jhovold: oh spiffy, I’ll pull those changes down.
<steev>
Would the rtc support also work for sdm845? Or is it specific to the 8280xp?
<jhovold>
the implementation is generic, but for the X13s I'm using a few PMIC registers to store the RTC time which may not be available on the sdm845
<jhovold>
If that machine uses UEFI, you may be able to use the UEFI RTC support after just enabling it in the dts
<jhovold>
so either you need to find some non-volatile storag (and nvmem cell) and add that to the dts, or
<jhovold>
you define the rtc nodes and specify 'qcom,uefi-rtc-info'
<jhovold>
well, actually the driver will not probe if that variable is missing so just adding that property does not hurt
alfredo has joined #aarch64-laptops
<juergh>
abelvesa, Looks good, UBSAN seems happy now. Is there anything else that changes with this patch that should be noticeable? I'll reply to the email as well. thanks for the quick turn around.
<abelvesa>
juergh: seems I messed up something, so I need to send v2 soon