On Tue, 6 Feb 2024 at 18:28, Ivan Semenov <ivan@xxxxxxxxxxx> wrote: > > The Linux mmc driver currently encounters a problem with eMMC chips connected via a 1-bit bus, when the chip supports a 4-bit bus. This regression was introduced in commit 577fb13 after functioning correctly in kernel version 3.15. > > In the function mmc_select_bus_width, the driver attempts to switch to 8-bit and 4-bit modes, but if the 4-bit bus test fails, it does not fall back to 1-bit mode and leaves eMMC in broken 4-bit mode. This results in I/O errors and failure to read the partition table. > > This patch addresses the issue by ensuring that the driver fallback to 1-bit bus mode if the attempt to switch to 4-bit mode fails. > > dmesg log for Samsung eMMC 5.1 chip connected via 1bit bus (only D0 pin) before patch: Wow, that was an old bug you found there. Just to make sure I understand correctly, the platform only supports 1-bit bus, but the DTS doesn't reflect that correctly as it instead indicates that 4 and 8-bit modes are supported? Don't get me wrong, I think the initialization-error-path should be able to cope with that, so we should certainly fix it! > > [134509.044225] mmc0: switch to bus width 4 failed > [134509.044509] mmc0: new high speed MMC card at address 0001 > [134509.054594] mmcblk0: mmc0:0001 BGUF4R 29.1 GiB > [134509.281602] mmc0: switch to bus width 4 failed > [134509.282638] I/O error, dev mmcblk0, sector 0 op 0x0:(READ) flags 0x0 phys_seg 1 prio class 2 > [134509.282657] Buffer I/O error on dev mmcblk0, logical block 0, async page read > [134509.284598] I/O error, dev mmcblk0, sector 0 op 0x0:(READ) flags 0x0 phys_seg 1 prio class 2 > [134509.284602] Buffer I/O error on dev mmcblk0, logical block 0, async page read > [134509.284609] ldm_validate_partition_table(): Disk read failed. > [134509.286495] I/O error, dev mmcblk0, sector 0 op 0x0:(READ) flags 0x0 phys_seg 1 prio class 2 > [134509.286500] Buffer I/O error on dev mmcblk0, logical block 0, async page read > [134509.288303] I/O error, dev mmcblk0, sector 0 op 0x0:(READ) flags 0x0 phys_seg 1 prio class 2 > [134509.288308] Buffer I/O error on dev mmcblk0, logical block 0, async page read > [134509.289540] I/O error, dev mmcblk0, sector 0 op 0x0:(READ) flags 0x0 phys_seg 1 prio class 2 > [134509.289544] Buffer I/O error on dev mmcblk0, logical block 0, async page read > [134509.289553] mmcblk0: unable to read partition table > [134509.289728] mmcblk0boot0: mmc0:0001 BGUF4R 31.9 MiB > [134509.290283] mmcblk0boot1: mmc0:0001 BGUF4R 31.9 MiB > [134509.294577] I/O error, dev mmcblk0, sector 0 op 0x0:(READ) flags 0x80700 phys_seg 1 prio class 2 > [134509.295835] I/O error, dev mmcblk0, sector 0 op 0x0:(READ) flags 0x0 phys_seg 1 prio class 2 > [134509.295841] Buffer I/O error on dev mmcblk0, logical block 0, async page read > > After patch: > > [134551.089613] mmc0: switch to bus width 4 failed > [134551.090377] mmc0: new high speed MMC card at address 0001 > [134551.102271] mmcblk0: mmc0:0001 BGUF4R 29.1 GiB > [134551.113365] mmcblk0: p1 p2 p3 p4 p5 p6 p7 p8 p9 p10 p11 p12 p13 p14 p15 p16 p17 p18 p19 p20 p21 > [134551.114262] mmcblk0boot0: mmc0:0001 BGUF4R 31.9 MiB > [134551.114925] mmcblk0boot1: mmc0:0001 BGUF4R 31.9 MiB To allow me to apply the patch, you need to provide your Signed-off-by-tag. According to the below: Signed-off-by: Ivan Semenov <ivan@xxxxxxxxxxx> Please tell me if you are okay with that, then I can amend the patch when applying. No need for you to post a new version. > --- > drivers/mmc/core/mmc.c | 2 ++ > 1 file changed, 2 insertions(+) > > diff --git a/drivers/mmc/core/mmc.c b/drivers/mmc/core/mmc.c > index f410bee50132..58ed7193a3ca 100644 > --- a/drivers/mmc/core/mmc.c > +++ b/drivers/mmc/core/mmc.c > @@ -1015,10 +1015,12 @@ static int mmc_select_bus_width(struct mmc_card *card) > static unsigned ext_csd_bits[] = { > EXT_CSD_BUS_WIDTH_8, > EXT_CSD_BUS_WIDTH_4, > + EXT_CSD_BUS_WIDTH_1, > }; > static unsigned bus_widths[] = { > MMC_BUS_WIDTH_8, > MMC_BUS_WIDTH_4, > + MMC_BUS_WIDTH_1, > }; > struct mmc_host *host = card->host; > unsigned idx, bus_width = 0; For my understanding, does your platform support HS200 mode too? Or only high-speed mode? The reason for my question is that it would be interesting to understand whether we end up running the error path in mmc_select_hs200() or not. [...] Kind regards Uffe