05:37 <chewitt> @xdarklight: from an old email with someone at Amlogic, the HDMI IP in the Meson 6 (not 8) was Transwitch

05:37 <chewitt> I will ask about Meson 8

05:37 <chewitt> but maybe they didn't switch to Synopsys until GXBB

05:39 <chewitt> Transwitch acquired by Cadence in 2014 .. also San Jose based so in the same neighbourhood as Amlogic

05:39 <xdarklight> chewitt: oh, thank you! I believe it's the same IP all the way back to Meson1

05:41 <xdarklight> "On November 21, 2013, TranSwitch Corporation filed a Chapter 7 petition for liquidation."

05:43 <chewitt> the Candence office is a 7 min drive from Amlogic's (which I've been to)

05:43 <chewitt> Cadence

05:43 <chewitt> https://ip.cadence.com/news/468/330/Cadence-Acquires-High-Speed-Interface-IP-Assets-of-TranSwitch-Corporation-Further-Expanding-IP-Portfolio-for-Mobile-Consumer-Market

05:50 <xdarklight> so that'll be my TODO for tonight. daytime job starts now. thanks again chewitt and see you later!

05:51 <chewitt> what info do you need on it?

05:51 <chewitt> target reconnaisance is one of my special skills :)

05:53 <xdarklight> a lot... my first question is the register split

05:54 <xdarklight> there's a part called "TX registers" which are obviously for the HDMI TX controller. but then there's something called "OTHER base registers" which control the interrupts

05:55 <xdarklight> but I don't know if all of this is from the TranSwitch IP or if that "OTHER register base" is similar to the TOP registers on GX which "extend" the DW HDMI IP

08:03 <narmstrong> xdarklight: the gx AO-CEC is the CEC hdl extract from the Transwitch HDMI controller, it was confirmed by Yixun, I forgot this until chewitt gave back the Transwitch name

08:03 <narmstrong> so yeah it's a Transwitch HDMI controller

08:16 <narmstrong> xdarklight: must be this one https://www.chipestimate.com/log.php?from=%2FTranswitch%2FTXC-48353-HDMI-14-Rx-with-HEAC-at-35Gbps%2Fdatasheet%2Fip%2F21436&logerr=1

08:16 <narmstrong> "HD-PXL-1.4"

08:17 <narmstrong> and maybe https://www.chipestimate.com/log.php?from=%2Fip.php%3Fid%3D17273&logerr=1 for older SoCs

09:11 <chewitt> datasheets aren't downloadable

09:12 <chewitt> I think Yixun left Amlogic, we had him in Slack for a while, but he's deactivated now

11:46 <Darkmatter66> narmstrong, Hi neil I did some testing for the screen hot-plug issue

11:47 <Darkmatter66> after powering the board(while the screen is plugged in and powered on):

11:48 <Darkmatter66> http://0x0.st/ijH9.txt

11:48 <Darkmatter66> modetest: http://0x0.st/ijHO.txt

11:49 <Darkmatter66> After powering off the screen from the remote(and while the screen is off): the outputs are the same from both the state file and modetest

11:49 <Darkmatter66> after powering on the screen from remote I can't cat the state file or use modetest while it's trying the outputs(while printing "flip_done timed out" lines)

11:50 <Darkmatter66> trying to cat the state file or running modetest just hangs until those lines are printined and the screen box has output on the HDMI connector

11:51 <Darkmatter66> after the screen has singal again, the state file output doesn't change

11:51 <Darkmatter66> for modetest this is the output: http://0x0.st/ijH4.txt

11:52 <Darkmatter66> Note that the modetest shows that Composite-1 is connected but I don't have anything connected to the board through the composite port so maybe this is the problem?

11:52 <Darkmatter66> I don't know if "connected" means there is something connected to the port or simply the port exist on the board and described in the device tree ?

11:59 <Darkmatter66> narmstrong, hmmmm, I disabled the cvbs-connector in device tree and it seems to fix/workaround the problem!

12:00 <Darkmatter66> So the driver thinks there is something connected to cvbs and that's why it's crashing maybe ?

12:45 <narmstrong> Darkmatter66: there is no way to say if cvbs is connected or not, so it's always set as Connected

12:45 <narmstrong> so DRM fallsback to cvbs if HDMI is disconnected

12:46 <narmstrong> it's expected

12:48 <narmstrong> Darkmatter66: I was told to switch to https://termbin.com/a01y

12:48 <narmstrong> Darkmatter66: could check the behavior with this ?

12:48 <narmstrong> since CVBS is not really used but the connector exists and works, I can't really disable it

16:49 <xdarklight> narmstrong: that's very like it then. only your link is for the HDMI RX, but there's a HDMI TX with "HD-PXL-1.4" and part number TXC-48352 (yours ends with 3): https://www.chipestimate.com/log.php?from=%2FTranswitch%2FTXC-48352-HDMI-14-Tx-with-HEAC-at-35Gbps%2Fdatasheet%2Fip%2F21435&logerr=1

16:50 <xdarklight> chewitt: while trying some google magic to find anything about the relation between Amlogic and TranSwitch I found that you told me in 2018 (two years ago) that Meson6 is a TranSwitch part. either I missed that or forgot about this, sorry :-(

18:06 <armoon> xdarklight-linux I was able to bootup the your latest kernel patches using eMMC module on C1+ thanks ;D (y) , I will give a try with HDMI testing at my end and let you know how it works, Thanks

18:06 <xdarklight> armoon: great! can you please reply to "Amlogic 32-bit Meson SoC SDHC MMC controller driver" (v5) with your Tested-by ?

18:07 <xdarklight> (sent on Mar 28, I'll ping the MMC maintainer tomorrow or on Friday since I haven't heard any update from him)

18:07 <armoon> yes I will submit that

18:18 <armoon> done thanks,

18:19 <xdarklight> awesome, thank you :)

18:20 <xdarklight> now there's even three Tested-by!

18:24 <xdarklight> narmstrong: it seems like Amlogic not only used the CEC part. HDMITX_TOP_HPD_FILTER in meson_dw_hdmi also looks a lot like the TranSwitch one which sets 0xa0 and 0xa0 (the GX driver does the same, but it's expressed as ((0xa0 is >> 4) << 12) instead of: (0xa0 << 8) ) =)

18:31 <narmstrong> xdarklight: interesting, since dw-hdmi doesn’t support hpd (it’s on the phy) they kept it

18:32 <narmstrong> They maybe kept part on the PHY aswell

18:34 <xdarklight> honestly, I wouldn't be surprised if they kept the PHY

18:47 <xdarklight> but that would mean that the development of G12A was interesting for the team which maintains that PHY, since they had to add support for 16b/18b coding

18:48 <Darkmatter66_> narmstrong, I'm recompiling with your patch for testing right now, but it started to recompile the whole thing, Does anyone know why

18:49 <Darkmatter66_> usually if no config or .c file has changed it doesn't recompile, if i change a .c file it only recompiles the module with that specific .c file

18:50 <Darkmatter66_> now every time i tell it to recompile it recompiles the whole thing

19:55 <Darkmatter66> narmstrong, aaand nope, your patch doesn't fix it :(