From 55e3b2f45880faaf06f3c660ca9e8a6d9aa14bce Mon Sep 17 00:00:00 2001 From: Kuninori Morimoto Date: Mon, 9 Dec 2019 15:29:52 +0900 Subject: wiki: Porting wiki: Porting Chat Log Signed-off-by: Kuninori Morimoto --- wiki/Chat_log/20181206-io-chatlog | 152 ++++++++++++++++++++++++++++++++++++++ 1 file changed, 152 insertions(+) create mode 100644 wiki/Chat_log/20181206-io-chatlog (limited to 'wiki/Chat_log/20181206-io-chatlog') diff --git a/wiki/Chat_log/20181206-io-chatlog b/wiki/Chat_log/20181206-io-chatlog new file mode 100644 index 0000000..e726f66 --- /dev/null +++ b/wiki/Chat_log/20181206-io-chatlog @@ -0,0 +1,152 @@ +09:04 < wsa_> ah, simon is excused +09:04 < wsa_> in which timezone is he now? +09:05 < wsa_> anyhow, let's start +09:05 < wsa_> welcome to this IO meeting +09:05 < wsa_> 1) status updates 2) topics +09:05 < geertu> wsa_: AFAIK, he's in California +09:05 < wsa_> Status updates +09:05 < wsa_> ============== +09:05 < wsa_> A - what have I done since last time +09:05 < wsa_> ------------------------------------ +09:05 < wsa_> Kaneko-san +09:05 < wsa_> : got I2C SYS-DMAC enablement for E3 merged +09:05 < wsa_> Marek +09:05 < wsa_> : improved the PCA953x driver to handle suspend/resume for SATA properly, +09:05 < wsa_> tested CAN on M3N using homemade CAN transceived, and bisected an upstream +09:05 < wsa_> network breakage +09:05 < wsa_> Niklas +09:05 < wsa_> : resent HS400 patch series about reset operation, ES revision handling, and +09:05 < wsa_> clock driver updates +09:05 < wsa_> Shimoda-san +09:05 < wsa_> : created patch and discussed proper upstream handling of some ethernet PHY +09:05 < wsa_> behaviour, handled the Q-Tag VLAN frame issue requested by the BSP team, +09:05 < wsa_> discussed SCIF flowchart issues, improved Gen3 USB2 phy handling, and +09:05 < wsa_> prepared a elinux page about USB virtualization +09:05 < wsa_> Simon +09:05 < wsa_> : got E3 IIC patches merged, sent patch for no-link property on Ebisu, and +09:05 < wsa_> provided feedback for the Q-Tag VLAN frame issue +09:05 < wsa_> Ulrich +09:05 < wsa_> : sent D3 CAN binding patches +09:05 < wsa_> Wolfram +09:05 < wsa_> : picked up I2C core PM work again, implemented patch and discussed about +09:05 < wsa_> WDT timer value during suspend, sent new version of MMC RPMB core fix and +09:05 < wsa_> improved some more smaller RPMB issues, answered GMSL questions from Jacopo, +09:05 < wsa_> reviewed IIC enablement, Niklas latest HS400 patches, and orchestrated +09:05 < wsa_> periupport some more +09:05 < wsa_> B - what I want to do until next time +09:05 < wsa_> ------------------------------------- +09:05 < wsa_> Geert +09:05 < wsa_> : wants to resubmit fixes for fallback to PIO in the sh-sci driver +09:05 < wsa_> Kaneko-san +09:05 < wsa_> : wants to enable PWM pn E3 +09:05 < wsa_> Niklas +09:05 < wsa_> : wants to keep tracking HS400 patches +09:05 < geertu> UTC-8? +09:06 < wsa_> Shimoda-san +09:06 < wsa_> : wants to continue to investigate USB2.0 host/peripheral resets behavior on +09:06 < wsa_> mainline, submit the USB virtualization page for eLinux, and wants to +09:06 < wsa_> investigate VFIO usage for Gen3 +09:06 < wsa_> Simon +09:06 < wsa_> : wants to follow up on the Q-Tag VLAN frame issue +09:06 < wsa_> Ulrich +09:06 < wsa_> : wants to send D3 CAN DT patches and do more patch review +09:06 < wsa_> Wolfram +09:06 < wsa_> : wants to submit I2C core PM work, check upstream SDHI bug report about +09:06 < wsa_> accessing large files, keep at watchdog and (older and revived) thermal +09:06 < wsa_> discussions, work on known IIC issues, more GMSL discussion +09:06 < wsa_> C - problems I currently have +09:06 < wsa_> ----------------------------- +09:06 < wsa_> Simon +09:06 < wsa_> : is waiting on various feedback about patches +09:07 < wsa_> so, my questions: +09:07 < wsa_> Marex: shimoda: are you in sync with upstreaming the PCIe ATF code? +09:08 < pinchartl> C - The pwm-rcar driver is broken with the PWM atomic API +09:08 < Marex> wsa_: I hope so, I need to rebase the PCI patch on top of upstream ATF, since they added all of the exception handling boilerplate and the patch doesn't apply anymore, but besides that, it should be doable +09:09 < wsa_> also, (not super much IO but upporting): who has a D3 board? According to the list, jmondi and Uli. Sound support needs to be upported... +09:09 < pinchartl> wsa_: I have Jacopo's D3 board, we exchanged the D3 and E3 +09:10 < shimoda> wsa_: about D3, I also have a board +09:10 < pinchartl> I thought I had updated the wiki, does it still show the D3 as being with Jacopo ? +09:10 < wsa_> pinchartl: will add the PWM issue to C), thanks! +09:10 < geertu> General comment: please update https://osdr.renesas.com/projects/linux-kernel-development/wiki/Hardware +09:11 < wsa_> pinchartl: you have. that was a race condition :D +09:11 < jmondi> ups, forgot to update this ^ +09:11 < jmondi> but somebody did already :) +09:12 < wsa_> so, if somebody is interested in that, there are some Jinso-patches for that on the list... +09:14 < wsa_> with simon not being here, that were all of my questions, I guess +09:14 < wsa_> are there some more from you, guys? +09:15 < pinchartl> if there's nothing else to discuss, I'd like to know if someone would like to address the PWM issue +09:15 < Marex> wsa_: speaking of D3/E3, I'm still curious about the HS400 performance on E3, it's lower than on M3N (with the same controller and same eMMC), both on BSP and mainline U-Boot +09:15 < geertu> Shimoda-san enquired about using VFIO. +09:15 < wsa_> shimoda: pinchartl: I assume you are linked now concerning the PWM issue? +09:15 < Marex> wsa_: I wonder if that's a QoS thing (if I fiddle with the QoS settings in ATF, it does affect that a bit) or something else +09:16 < neg> a cross IO/core question, do you think it's possible for us to aim to enable HS400 for 4.21 or are we a tad to late? +09:16 < pinchartl> shimoda: do you have time to work on pwm-rcar, or should I try to fix the problem ? +09:16 < wsa_> neg: I would very much hope so +09:17 < shimoda> pinchartl: i have time to work on pwm-rcar +09:17 < wsa_> neg: we could ask geertu if he is happy with the clock patches :) +09:17 < wsa_> neg: from the reviews, only minor stuff needs to be changed +09:18 < geertu> I plan to apply the clock patches, with the two nits fixed. +09:18 < shimoda> pinchartl: so I'd like to know how to reproduce the issue. +09:18 < geertu> And send a PR tomorrow. +09:18 < pinchartl> shimoda: great :-) I noticed the issue with backlight on E3, but I think it can be reproduced on any board. the problem comes from using the PWM atomic API. as the pwm-rcar driver doesn't suport it natively, the PWM core translates the calls to the legacy PWM API (set config, enable). the sequence of calls doesn't work well +09:18 < wsa_> geertu: one nit was not in the commit message but only in the changelog +09:19 < neg> wsa_: Yes even so I noticed Simon closed the renesas tree yesterday for 4.21 but maybe he can take it anyhow. Will ask him once he is awake +09:19 < geertu> wsa_: Which one? +09:19 < wsa_> s/rete/rate/ +09:19 < pinchartl> rcar_pwm_config() returns immediately because pwm->state.duty_cycle = 0 +09:20 < pinchartl> and then rcar_pwm_enable() will return an error because RCAR_PWMCNT_CYC0 == 0 and RCAR_PWMCNT_PH0 == 0 +09:20 < wsa_> neg: can you resend the other patch ASAP? +09:20 * geertu has a PWM déjà vue +09:20 < geertu> s/vue/vu/ +09:21 < pinchartl> so I think we need rcar_pwm_config() first compute the config parameters and store them, and then return if the PWM is not enabled, or program the hardware if it is +09:21 < wsa_> then let's hope simon will make an exception for the final HS400 DTS enablement patch +09:21 < pinchartl> then rcar_pwm_enable() should always proceed with the stored config +09:21 < pinchartl> and program the hardware if it's not programmed yet +09:21 < shimoda> pinchartl: I didn't know about the PWM atomic API. I'll check that. as you said, the driver prevents to enable it with CYC0 == 0 and PH0 == 0. +09:21 < geertu> Simon is still applying patches. +09:22 < neg> wsa_: I talked to geertu about the patches yesterday and he has generously agreed to fix the nit commetns when applying them. My question was more about if it was to late for the HS400 DT patch +09:22 < wsa_> neg: I don't think so :) +09:22 < neg> good :-) +09:23 < pinchartl> shimoda: it's a chicken and egg problem. rcar_pwm_config() doesn't program RCAR_PWMCNT because the channel is disabled, and then rcar_pwm_enable() errors out because RCAR_PWMCNT isn't programmed :-) +09:24 < geertu> Is this a recent PWM issue? +09:24 < pinchartl> one option to test this is to use my git://linuxtv.org/pinchartl/media.git drm/du/d3e3 branch +09:25 < pinchartl> make sure to compile the pwm-backlight driver +09:25 < pinchartl> then you can control the backlight from sysfs +09:25 < pinchartl> in /sys/class/backlight +09:25 < pinchartl> pwm-backlight now uses the PWM atomic API +09:25 < pinchartl> geertu: I don't think it's new, what has changed is the way client drivers use the PWM API +09:26 < pinchartl> the call sequence was always valid +09:26 < pinchartl> but never tested +09:26 < wsa_> eeeks, my thunderbird crashes badly...(?) so I can't look now... is there somebody at the ethernet breakage that Marex bisected? +09:26 < shimoda> pinchartl: :) so, i added a trickly code at line 157 of pwm-rcar.c. But, I think I should support .apply ops instead of legacy ops +09:27 < pinchartl> that would of course be a good option too :-) +09:27 < geertu> wsa_: It's hch's DMA cleanup +09:27 < geertu> Marex pointed to https://patchwork.kernel.org/patch/10668093/ +09:27 < wsa_> geertu: i see. thanks +09:28 < geertu> Aboiut Shimoda-san's VFIO question +09:28 < geertu> The answer is "yes, you can". SATA is the most mature example. You still need to apply out-of-tree kernel and qemu patches, cfr. https://elinux.org/R-Car/Virtualization/VFIO +09:28 < pinchartl> shimoda: the apply operation should be easier to implement for the driver, and should fix the probleem +09:29 < pinchartl> that's all on my side for pwm-rcar +09:29 < Marex> wsa_: the link also contains a patch , which fixes the breakage +09:29 < wsa_> cool +09:29 < pinchartl> on a different, common topic, I would like to talk about peripericon @FOSDEM2019. we can do that now or later as part of core or multimedia +09:30 < shimoda> pinchartl: thank you for the detailed information! I'll try to implement the apply operation. +09:30 < wsa_> Yeah, I'd like to talk about that if there are no more IO related questions? +09:30 < Marex> wsa_: speaking of D3/E3, I'm still curious about the HS400 performance on E3, it's lower than on M3N (with the same controller and same eMMC), both on BSP and mainline U-Boot +09:30 < Marex> wsa_: I wonder if that's a QoS thing (if I fiddle with the QoS settings in ATF, it does affect that a bit) or something else +09:30 < wsa_> shimoda: cool, thanks for taking care of that. +09:31 < shimoda> geertu: thank you aboit VFIO. I didn't check the eLinux page. I'll check it after I made a patch for pwm-rcar :) +09:31 < wsa_> Marex: did you try using the BSP also? +09:31 < geertu> shimoda: You're welcome. If you have issues, please ping me! +09:32 < Marex> wsa_: yes, that's why I wrote it happens both on BSP and mainline U-Boot +09:32 < Marex> wsa_: BSP Linux that is +09:32 < shimoda> geertu: sure! +09:33 < wsa_> Marex: ah linux, too, okay +09:34 < wsa_> Marex: If it happens the same on BSP and upstream, this is usually the point where I would ask BSP team or HW team... +09:34 < Marex> wsa_: I brought it up in the periperi list, but that discussion died out +09:34 < Marex> wsa_: I'll restart it +09:35 < wsa_> Marex: please do +09:36 < wsa_> okay +09:36 < wsa_> FOSDEM talk? +09:37 < wsa_> geertu: are you okay with that? +09:37 < geertu> Sure -- cgit v1.2.3