[discovery] - Android 8.1 - SD-Card support/reader broken

[Thaodan]

Platform: sony-nile Device: 8.1 /Mer-Hybris Kernel version: 4.4.148-g49441d2181d0 Android version: no /dev/mmcblk1 device

Description The xperia open devices page states that only sdcards that are <=32GB are supported. What is missing on SDXC kernel support? Is there some firmware or specific Kconfig needed?

The issue is that my device doesn't find the card and just throws a gpio error every time. log: https://paste.kde.org/pgcqwzuyp

Symptoms Write the symptoms here.

How to reproduce Write how to reproduce here.

[oshmoun]

Not sure if that gpio error is relevant, as it also appears on loire yet sdcard works. did you check the paths in fstab and whether they are actually valid? maybe the sdcard path needs to be changed. Also, is the card exFAT formatted?

[Thaodan]

The card isn't even detected by the controller. i have an Xperia x and here the card works. The card is formatted as ext4. I booting twrp which uses the stock kernel (kernel copyleft ) and there the card is detected just fine.

[Thaodan]

the dmesg.output there is :

<6>[ 0.790033] mmcc-sdm660 c8c0000.clock-controller: Registered MMSS clocks <6>[ 4.699441] sdhci_msm c0c4000.sdhci: No vmmc regulator found <6>[ 4.706826] sdhci_msm c0c4000.sdhci: No vqmmc regulator found <6>[ 4.712930] mmc0: SDHCI controller on c0c4000.sdhci [c0c4000.sdhci] using 64-bit ADMA in CMDQ mode <6>[ 4.808493] mmc0: Out-of-interrupt timeout is 50[ms] <6>[ 4.808532] mmc0: eMMC FW version: 0xa5 <6>[ 4.812539] mmc0: CMDQ supported: depth: 32 <6>[ 4.816084] mmc0: cache barrier support 1 flush policy 1 <6>[ 4.833413] mmc0: CMDQ enabled on card <6>[ 4.845182] mmc0: new HS400 MMC card at address 0001 <6>[ 4.863351] mmcblk0: mmc0:0001 HBG4a2 29.1 GiB <6>[ 4.873252] mmcblk0rpmb: mmc0:0001 HBG4a2 partition 3 4.00 MiB <6>[ 4.880435] mmcblk0: p1 p2 p3 p4 p5 p6 p7 p8 p9 p10 p11 p12 p13 p14 p15 p16 p17 p18 p19 p20 p21 p22 p23 p24 p25 p26 p27 p28 p29 p30 p31 p32 p33 p34 p35 p36 p37 p38 p39 p40 p41 p42 p43 p44 p45 p46 p47 p48 p49 p50 p51 p52 p53 p54 p55 p56 p57 p58 p59 p60 p61 p62 p63 p64 p65 p66 p67 p68 p69 p70 p71 p72 p73 p74 p75 p76 p77 p78 p79 <6>[ 4.926716] sdhci_msm c084000.sdhci: No vmmc regulator found <6>[ 4.926752] sdhci_msm c084000.sdhci: No vqmmc regulator found <6>[ 4.931929] mmc1: SDHCI controller on c084000.sdhci [c084000.sdhci] using 64-bit ADMA in legacy mode <6>[ 5.328872] mmc1: new ultra high speed SDR104 SDXC card at address aaaa <6>[ 5.332794] mmcblk1: mmc1:aaaa ACLCF 119 GiB <6>[ 5.409709] mmcblk1: p1

[jerpelea]

SDR104 SDXC card at address aaaa

<6>[ 5.332794] mmcblk1: mmc1:aaaa ACLCF 119 GiB <6>[ 5.409709] mmcblk1: p1 the card is detected by kernel

[Thaodan]

That was the message with the stock (kernel copyleft) kernel. The message with the open devices kernel (just kernel here on github) was:

[    0.000000] Kernel command line: rcupdate.rcu_expedited=1 core_ctl_disable_cpumask=0-7 lpm_levels.sleep_disabled=1 display_status=on selinux=0 androidboot.selinux=permissive msm_rtb.filter=0x3F ehci-hcd.park=3 coherent_pool=8M sched_enable_power_aware=1 user_debug=31 androidboot.hardware=discovery androidboot.verifiedbootstate=orange androidboot.keymaster=1  androidboot.veritymode=enforcing root=/dev/dm-0 dm="system none ro,0 1 android-verity /dev/mmcblk0p78" androidboot.bootdevice=c0c4000.sdhci cei_project_id=BY22 cei_phase_id=MP cei_sku_id=GINA cei_simslot_id=ss cei_mainboard_id=SM22 androidboot.ceibootmode=0 cei_color_id= androidboot.baseband=sdm mdss_mdp.panel=1:dsi:0:qcom,mdss_dsi_td4322_innolux_fhd_cmd:config0:1:none:cfg:single_dsi androidboot.slot_suffix=_a skip_initramfs rootwait ro init=/init  androidboot.bootloader=xboot oemandroidboot.xboot=1310-0301_X_Boot_SDM630_LA1.0_O_121 androidboot.serialno=CQ300049KL startup=0x00008000 warmboot=0x77665501 oemandroidboot.babe08b3=500000
[    0.345406] mmcc-sdm660 c8c0000.clock-controller: Registered MMSS clocks
[    1.425556] ## mmc0: gpio_is_valid(pdata->uim2_gpio)=-2: failure
[    1.428891] ## mmc0: can't set uim2_gpio: -2
[    1.429563] sdhci_msm c0c4000.sdhci: No vmmc regulator found
[    1.429640] sdhci_msm c0c4000.sdhci: No vqmmc regulator found
[    1.430344] mmc0: SDHCI controller on c0c4000.sdhci [c0c4000.sdhci] using 64-bit ADMA in CMDQ mode
[    1.514447] mmc0: Out-of-interrupt timeout is 50[ms]
[    1.514581] mmc0: eMMC FW version: 0xa5
[    1.514650] mmc0: CMDQ supported: depth: 32
[    1.514717] mmc0: cache barrier support 1 flush policy 1
[    1.523514] mmc0: CMDQ enabled on card
[    1.523646] mmc0: new HS400 MMC card at address 0001
[    1.524924] mmcblk0: mmc0:0001 HBG4a2 29.1 GiB
[    1.525308] mmcblk0rpmb: mmc0:0001 HBG4a2 partition 3 4.00 MiB
[    1.527569]  mmcblk0: p1 p2 p3 p4 p5 p6 p7 p8 p9 p10 p11 p12 p13 p14 p15 p16 p17 p18 p19 p20 p21 p22 p23 p24 p25 p26 p27 p28 p29 p30 p31 p32 p33 p34 p35 p36 p37 p38 p39 p40 p41 p42 p43 p44 p45 p46 p47 p48 p49 p50 p51 p52 p53 p54 p55 p56 p57 p58 p59 p60 p61 p62 p63 p64 p65 p66 p67 p68 p69 p70 p71 p72 p73 p74 p75 p76 p77 p78 p79
[    1.538583] ## mmc1: gpio_is_valid(pdata->uim2_gpio)=-2: failure
[    1.552056] ## mmc1: can't set uim2_gpio: -2
[    1.553454] sdhci_msm c084000.sdhci: No vmmc regulator found
[    1.553530] sdhci_msm c084000.sdhci: No vqmmc regulator found
[    1.554124] mmc1: SDHCI controller on c084000.sdhci [c084000.sdhci] using 64-bit ADMA in legacy mode
[    4.076078] EXT4-fs (mmcblk0p78): mounted filesystem with ordered data mode. Opts: barrier=1
[    4.102156] EXT4-fs (mmcblk0p70): mounted filesystem with ordered data mode. Opts: barrier=1
[    4.102804] EXT4-fs (mmcblk0p79): mounted filesystem with ordered data mode. Opts: (null)
[    4.119289] EXT4-fs (mmcblk0p68): mounted filesystem with ordered data mode. Opts: barrier=1
[    4.122434] EXT4-fs (mmcblk0p44): mounted filesystem with ordered data mode. Opts: barrier=1,data=ordered,nodelalloc,errors=panic
[    4.134434] EXT4-fs (mmcblk0p2): warning: maximal mount count reached, running e2fsck is recommended
[    4.135431] EXT4-fs (mmcblk0p2): mounted filesystem with ordered data mode. Opts: barrier=1,data=ordered,nodelalloc,errors=panic
[    5.711334] droid-hal-init: Opened file '/d/mmc0/mmc0:0001/ext_csd', flags 0

No device file is created and no entry in sysfs.

[jerpelea]

the sdfat driver was added and configured as default https://github.com/ab123321/kernel/blob/7a4f7261a10169ca9cfdaaae4b6b93b0cdcc3893/fs/sdfat/Kconfig#L3

[Thaodan]

Is that even needed to detect the block file on the sdcard? My card is formated as ext4 and I even can't mount the file. To what are the gpio errors related to?

[jerpelea]

I am sorry but I missed the ext4 format no in this case is just the sdcard driver that needs some updates from copylefts can you diff the 2?

[Thaodan]

sure, bur which parts? I think the changes are to large for a full diff. Where is the driver specific to the controller?

[jerpelea]

I would start from kernel/drivers/mmc/host

[Thaodan]

which parts of that are specific to the soc of nile?

[Thaodan]

I think I've found the error line that triggers the message: https://github.com/mer-hybris/android_kernel_sony_msm/blob/hybris-sony-aosp-8.1.0_r35_20180714/drivers/mmc/host/sdhci-msm.c#L1847

[Thaodan]

Tested a 16GB(SDHC) card same error.

[Thaodan]

Any help on this?

[oshmoun]

did you confirm fstab is ok? maybe paths simply need adjusting

[Thaodan]

fstab is ok. Even when the fstab is wrong the card should be visible in dmesg?

[Thaodan]

Added log with CONFIG_MMC_DEBUG=y compiled in. dmesg.log

One thing I noticed was that the internal mmc gets re scanned several times but no the external. Is that intended?

[jerpelea]

I tested pioneer with 4.9 kernel on Android 9 and a 32Gb card pioneer:/ $ ls /dev/block/mmcblk1 /dev/block/mmcblk1

[Thaodan]

So the bug is only on discovery? Any further advice to debug this?

[jerpelea]

I will check the same device with 8.1 and kernel 4.4 then we can look at discovery

[dimitar10]

I tested with 128GB ext4 formatted sdcard on Pioneer, with latest 4.4 prebuilt kernel and latest AOSP 8.1, and there's a format unsupported error. If the card is formatted within the device (to fat), everything works OK. I'm attaching the dmesg if it can be of any help.

sdcard-128GB-ext4-not-recognized-pioneer-dmesg.log

screenshot.zip

[Thaodan]

I will check the same device with 8.1 and kernel 4.4 then we can look at discovery

That works. I have someone here that tested it with bot 8GB and 64GB card (pioneer).

[jerpelea]

lets not mix the things we have to check first that the regular 32Gb card is detected

[Thaodan]

@dimitar10 have tried mounting the card by hand?

FAT-fs (mmcblk1p1): Volume was not properly unmounted. Some data may be corrupt. Please run fsck. Looks suspicions.

[jerpelea]

I tested Discovery on Android 9 and kernel 4.9 and the 32Gb sdcard is not detected

[Thaodan]

we have to check firt that the regular 32Gb card is detected.

On pioneer?

[Thaodan]

I tested Discovery on Android 9 and kernel 4.9 and the 32Gb sdcard is not detected

Can you build the pioneer kernel with MMC_DEBUG=y, that we have something to compare?

How is the card connected to the phone?

[jerpelea]

tests completed pioneer kernel 4.4 - OK kernel 4.9 - OK

discovery kernel 4.4 - NOK kernel 4.9 - NOK

[jerpelea]

@Thaodan I recommend looking here for a quirk then backport it (drivers/mmc/host/sdhci-msm.c) https://github.com/sonyxperiadev/kernel-copyleft/commit/ba16212af72a789c776900d35ba1483cce4e08d3

[Thaodan]

Ported that quirk back. Is there a more detailed explanation why this boards need the workaround? EDIT: is their an upstream version of cei?

[Thaodan]

I applied the quirk by using CONFIG_MACH_SONY_DISCOVERY instead of cei. But still no sdcard found. Should I have applied less of the patch or do I miss something?

https://github.com/Thaodan/android_kernel_sony_msm/commit/6d6fac80958af76e0091c67d9cbbc37fb9a5ffc9

dmesg|grep mmc1 whith CONFIG_MMC_DEBUG=y: https://paste.kde.org/penjpntqf#line-42

[Thaodan]

full dmesg dmesg.log

It looks like the initialisation of the mmcblk1 goes further than without the patch but after CMD1 it has 0 Hz.

[Thaodan]

@jerpelea did I used the right workaround? Is it possible to get more deeper information on why the workaround and how it was handled in the stock kernel?

[jerpelea]

I linked the stock kernel diff. Unfortunately, for the moment, I do not have time to look deeper at this issue

[Thaodan]

That's why I asked if I used the right part of the workaround. It feels like I miss something. Its bogus that such part is broken in this device but not in the sister device.

[jerpelea]

fix found and pushed https://github.com/sonyxperiadev/kernel/commit/103888a2221c4fa824ed29f5b6fe2d0326042a03

[jerpelea]

I tested both discovery and voyager and both are OK

[Thaodan]

Thanks! can you explain why this board is different in this regard compared to nile?

[Thaodan]

Can you merge this to 4.4.x too?

[jerpelea]

it is already merged and I will push the patch ASAP

[Thaodan]

Was already listed in arch/arm/boot/dts/qcom/sdm630-qrd.dtsi#L249 btw.

[jerpelea]

that is not your device !

[Thaodan]

I know but its the base/reference board device tree supplied by quallcomm.

[jerpelea]

the device tree contains several reference motherboards built on the same platform while nile a new board.

[ghost]

@Thaodan does TWRP work for you with unmodified open devices AOSP or do you have to use a different kernel, compile framdisk etc.? I'm asking because of this: https://twrp.me/sony/sonyxperiax.html (look at Notes About Sony Devices)

[Thaodan]

It did work for me with the twrp image posted on xda, it uses the kernel from the stock rom. However the open devices kernel should work now too. Everything else isn't changed.