05:48 <ndec> oh, that's great! (pushing these things into debian!)

05:55 <amstan> steev: what's pd-mapper?

10:24 <mani_s> steev, updated the lenovo-x13s branch based on Johan's tree (wip/sc8280xp-next-20220728)

10:25 <mani_s> this should have working suspend/resume, trackpad on most of the variants, external display etc...

10:25 <mani_s> there is still no battery driver update to query the battery status. But this should be fixed very soon

10:27 <mani_s> as someone noted issue, we had to drop the gic-its patch that gives the affinity support to MSIs. this is being investigated and hopefully fixed soon

10:27 <mani_s> well i do have a workaround for that but that will decrease the speed of wifi ;)

10:29 <maz> mani_s: what's the issue with the ITS?

10:30 <mani_s> maz, if i use its for handling msi's, i'm seeing pcie aer correctable errors from one of the pcie controllers

10:30 <mani_s> if i limit that controller's link speed to gen2, then the aer errors are gone

10:30 <maz> mani_s: do you have an SMMU on this path?

10:30 <mani_s> maz, yes

10:30 <maz> huh.

10:31 <maz> and you only see the errors when targeting the ITS, and not for general DMA?

10:31 <mani_s> maz, i'm not sure how aer is getting triggered with its

10:31 <mani_s> maz, with dwc internal msi controller, i don't see aer errors

10:32 <maz> I'm not sure how the GEN2 stuff would impact the MSI transactions, unless this triggers some other bug in the PCIe IP.

10:32 <mani_s> also, i verified that aer interrupt is mapped correctly with dwc msi also. tried injecting the aer errors manually and I saw the error logs

10:32 <mani_s> so with dwc msi, aer irq is working fine but there are actually no errors

10:32 <maz> worth talking to synopsis about that...

10:33 <mani_s> maz, yeah, i'm talking to the qcom hardware guys about this

10:34 <mani_s> maz, the fact that limiting gen speed silences the aer errors makes sense. because, it could some issue with the board design so operating at the gen3 speed could be broken for a specific controller

10:35 <mani_s> but i'm puzzled why the error was not coming with dwc msi

10:36 <maz> mani_s: but I'd expect that if there were some signal integrity issues, they'd also show up for most DMA operations (MSIs are extremely rare compared to the DMA traffic).

10:36 <maz> some something doesn't add up here.

10:37 <mani_s> maz, good point

10:39 <maz> which makes me think that this is a GEN3 issue in combination with external MSIs that trigger the problem, and that'd definitely be a RC thing.

10:39 <mani_s> maz, agree

10:41 <mani_s> maz, but one thing to note here is, MSI is not the one triggering aer errors. once its is used for handling the msi's then the issue appears

10:42 <mani_s> but i do think the issue belongs to the pcie ip

10:43 <mani_s> initially I thought that with dwc msi, there could be some issue with aer irq handling so the errors were lost... but when i tried injecting the aer errors manually on dwc msi setup, i could see the errors

10:43 <maz> ISTR that the DWC IP has some special settings to forward MSI transactions outside of the RC.

10:44 <maz> anyway, this really is something QC and Synopsys should clarify.

10:45 <mani_s> maz, but everything is working fine for other pcie instances on the same soc. only one instance is the issue

10:45 <mani_s> maz, sure

10:45 <maz> it'd also be interesting to see if you have the same issue when targetting the MBI register in the distributor instead the ITS TRANSLATER.

10:46 <maz> I'm sure you have a small range of spare SPIs for that.

10:47 <mani_s> maz, interesting... i'll try that

13:00 <steev> mani_s: oh awesome

13:10 <steev> mani_s: i think this weekend i want to try to go over it all and replace the stuff that is in there with what all got submitted over the past few days, a lot hit the mailing list :)

17:36 <clover[m]> <mani_s> "this should have working suspend..." <- pog

18:33 <clover[m]> another thermal driver update available on x13s from Commercial Vantage

18:48 <steev> i wonder if we could get that from the fwupdate service - it's not specific to the x13s, according to the changelog