On Wed, Jun 08, 2022 at 01:09:54AM +0800, Yongqin Liu wrote: > Hi, All > > With the 4.14.281 version[1], there were three mmc related changes merged, > but that causes one boot failure with the X15 Android builds, a problem > similar to one reported before here[2]. > After being confirmed with Ulf Hansson, and verified with the X15 Android build, > it needs to have the following four commits cherry-picked to the 4.14 > branch as well. > > 4f32b45c9a2c mmc: core: Allow host controllers to require R1B for CMD6 > 5fc615c1e3eb mmc: core: Respect MMC_CAP_NEED_RSP_BUSY for erase/trim/discard > d091259b8d7a mmc: core: Respect MMC_CAP_NEED_RSP_BUSY for eMMC sleep command > 23161bed631a mmc: sdhci-omap: Fix busy detection by enabling > MMC_CAP_NEED_RSP_BUSY > > The above four commits are from the 4.19 branch, as they are a little > easier to be cherry-picked > into the 4.14 branch, compared to the commits from the mainline branch. > (I have confirmed that the four commits are all in 4.19, 5.4, 5.10 and > mainline branches already). > > Saying that, there will be still one merge conflict reported when > cherry picking the commit of > 4f32b45c9a2c, it's easy to resolve though. > To avoid the merge conflict, it could be done like this as well: > 1. revert the 327b6689898b commit from 4.14 first, so that the commits in step#2 > could be cherry-picked without any problem > 327b6689898b mmc: core: Default to generic_cmd6_time as > timeout in __mmc_switch() > 2. git cherry-pick the following commits from 4.19 into the 4.14 branch > 4f32b45c9a2c mmc: core: Allow host controllers to require R1B for CMD6 > 5fc615c1e3eb mmc: core: Respect MMC_CAP_NEED_RSP_BUSY for > erase/trim/discard > d091259b8d7a mmc: core: Respect MMC_CAP_NEED_RSP_BUSY for eMMC > sleep command > 23161bed631a mmc: sdhci-omap: Fix busy detection by enabling > MMC_CAP_NEED_RSP_BUSY > 26c6f614cf02 mmc: mmc: core: Default to generic_cmd6_time as > timeout in __mmc_switch() > The last commit of 26c6f614cf02 is for the revert in step#1. > > I am not sure which way is more convenient for the maintenance work > here, so just list both of them here > for your information. > And please let me know if there is anything else I could help on this > cherry pick work here. Please send properly backported patches to us, trying to do the revert and fixup like you describe above is going to be hard to verify I got it right. A series of patches is best as that way we know you tested it properly and sent us the correct patches. thanks, greg k-h
