07:38 <arnd> drewfustini: dw_mmc.c is designware's own MMC controller design with a custom register layout, which used to be fairly popular, but I'm guessing that it is no longer updated for the latest specifications

07:39 <arnd> SDHCI is a vendor-neutral spec for SD/MMC controllers with a standard register layout implemented by a number of companies including designware, which licenses it to other parties

07:41 <arnd> I would guess that a number of the other sdhci-*.c files are actually designware based as well, but use the common sdhci code directly rather than the intermediate sdhci-of-dwcmshc.c abstraction. Hard to know for sure

07:43 <arnd> I can see that the original contribution of sdhci-of-dwcmshc.c was done by Jisheng Zhang, who maintains the Synaptics (formerly Marvell, not related to Synopsys) Berlin platform, but he never mainlined the later Berlin dts files

07:43 <geertu> There's also SDHI and TMIO

07:46 <geertu> And given the tendency to have NDA'd documentation for SD-card controllers, it's difficult to know who's using what, and what it is derived from.

07:47 <geertu> Still, SD managed to cry victory over Sony MS ;-)

10:23 <Amit_T> External aborts on ARM64 can be synchronous as well asynchronous, but if EA's are signaled through SError, are these always asynchronous ?

10:56 <Xogium> so, there any way to confirm my theory that its because linux console subsystem doesn't like having congested console if there are multiple ones connected to printk/dmesg or what have you ?

10:57 <Xogium> really curious to determine if that was really the issue I hit, with somehow putting ttyGS0 before S0 made every boot issue disappear, the hang, the reboot in your face unless you actually listen to the ttyACM port

11:00 <Xogium> weirdest issue I ever hit in a long while, the last one that was that insane was somehow managing to cause a kernel panic and still have userspace running

13:37 <drewfustini> arnd and geertu: thank you for much for your insights

13:53 <drewfustini> arnd: I didn't realize that Synaptics bought part of Marvell. I have to admit "berlin" always catches my eye when grepping through linux

13:53 <drewfustini> I appreciate knowing the history

14:06 <arnd> drewfustini: individual business units get moved around a lot. Synaptics also has Broadcom's wireless SoCs, NXP has Marvell's wireless SoCs and Marvell has Broadcom's server chips (by way of Cavium, but then discontinued)

16:20 <maz_> Amit_T: no, see SCR_EL3.EASE=1.

16:41 <Amit_T> maz_: okay, I see this SCR_EL3.EASE=0, so its pretty much an asynchronous one, right?

16:47 <Amit_T> sorry, taking the above statement back. its the way its routed but always synchronous.

18:32 <drewfustini> arnd: thanks for the advice about SDHCI. I see now in the databook that "DWC_mshc_crypto implements all the registers defined by the SD Host Controller Specification that defines registers for SD, SDIO and UHS-II modes. Some of these registers are reused in the eMMC mode." This SoC has that IP configured for eMMC only card interface type. I am hoping that means the SDHCI driver will still apply.

18:34 <drewfustini> The probe fails about half way through on a timeout so I'm going slowly going through all the MMIO read and writes to see what is happening with the various registers