> -----Original Message----- > From: Marcel Ziswiler <marcel.ziswiler@xxxxxxxxxxx> > Sent: 2022年11月3日 4:43 > To: linux-arm-kernel@xxxxxxxxxxxxxxxxxxx; linux-mmc@xxxxxxxxxxxxxxx; > linux-kernel@xxxxxxxxxxxxxxx > Cc: dl-linux-imx <linux-imx@xxxxxxx>; kernel@xxxxxxxxxxxxxx; > s.hauer@xxxxxxxxxxxxxx; Bough Chen <haibo.chen@xxxxxxx>; > ulf.hansson@xxxxxxxxxx; shawnguo@xxxxxxxxxx; adrian.hunter@xxxxxxxxx; > festevam@xxxxxxxxx > Subject: mmc: host: sdhci-esdhc-imx: async sdio interrupt and hardware > auto-tuning > > Hi there > > When we investigated an SDIO Wi-Fi stability issue I discovered the following > downstream-only commit in NXP's latest downstream BSP > 5.15.52_2.1.0 [1]. The asynchronous SDIO interrupt seems to adversly affect the > hardware auto-tuning. As a workaround what got implemented is an fsl-custom > device tree property that limits the auto-tuning to > 1 bit only (e.g. on DAT[0]) while the SDIO interrupt stays on DAT[1]. > > Does anybody know any further details about this? > > Would this be a reasonable thing to also implement upstream? Hi Marcel, I send these patches to community few month ago, get some comments related to dts part. I will continue to do the upstream work for these patches. Best Regards Haibo Chen > > Unfortunately, so far we have not really done any extensive SDIO validation > using upstream but I plan to set something like that up so we may see whether > or not we are also hitting such issue. > > Thanks! > > [1] > https://eur01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.co > m%2Fnxp-imx%2Flinux-imx%2Fcommit%2F84c529a1474a9f6614e475e178ca5fe > 23b6177de&data=05%7C01%7Chaibo.chen%40nxp.com%7C5b75121c05a > d48c362d708dabd12e507%7C686ea1d3bc2b4c6fa92cd99c5c301635%7C0%7C0 > %7C638030186093684545%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAw > MDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7 > C&sdata=PX17UiGkQHfMxHd9Sf2jEbIsZgWO5MCp5pF%2BIDXQIn4%3D&a > mp;reserved=0 > > Cheers > > Marcel
