| 1 | This patch makes SDHC cards work with the mmc_spi driver. |
| 2 | |
| 3 | The problem is that they fail when reading the last block of the card using |
| 4 | a multi-block read. This is because on SDHC the multiple block read has to be |
| 5 | stopped with an explicit STOP command, which needs to be sent to the card |
| 6 | while the incoming transfer is in progress. |
| 7 | The 2.6.3[45] mmc-spi driver sends it after the last block transfer, so the |
| 8 | SDHC card continues reading past the end of the card. |
| 9 | This patch works around this by using single-block reads if we're reading the |
| 10 | last blocks of the card. |
| 11 | -martinwguy, 14 May 2010 |
| 12 | |
| 13 | Date: Thu, 29 Apr 2010 21:30:36 +0300 |
| 14 | From: Mika Westerberg <mika.westerberg@iki.fi> |
| 15 | To: Martin Guy <martinwguy@gmail.com> |
| 16 | |
| 17 | On Wed, Apr 21, 2010 at 02:10:08AM +0100, Martin Guy wrote: |
| 18 | > |
| 19 | > the SDHC cards I have don't work at all, spewing tons of: |
| 20 | > mmcblk0: error -38 sending status comand |
| 21 | > mmcblk0: error -38 sending read/write command, response 0x4, card status 0xff04 |
| 22 | > end_request: I/O error, dev mmcblk0, sector 7744509 |
| 23 | |
| 24 | I bought today a new 4GB SDHC card and with that I get similar |
| 25 | errors that you are getting. I hacked around quick fix which seems |
| 26 | to work in my case. I'm wondering whether you could check if it |
| 27 | helps with your SDHC card as well? |
| 28 | |
| 29 | This problem is easy to reproduce, just read last sector of the |
| 30 | card (I wrote simple C program but running fdisk -l does the same). |
| 31 | |
| 32 | Patch is below. |
| 33 | |
| 34 | Thanks, |
| 35 | MW |
| 36 | |
| 37 | From: Mika Westerberg <mika.westerberg@iki.fi> |
| 38 | Date: Thu, 29 Apr 2010 21:14:32 +0300 |
| 39 | Subject: [PATCH] mmc_block: use single block reads for last block on SPI |
| 40 | |
| 41 | Some SD-cards fail when doing multiblock read for last block with SPI host. Real |
| 42 | reason is not known but as workaround we can perform this last read using |
| 43 | multiple single block reads. |
| 44 | |
| 45 | Signed-off-by: Mika Westerberg <mika.westerberg@iki.fi> |
| 46 | --- |
| 47 | drivers/mmc/card/block.c | 19 +++++++++++++++++++ |
| 48 | 1 files changed, 19 insertions(+), 0 deletions(-) |
| 49 | |
| 50 | --- a/drivers/mmc/card/block.c |
| 51 | +++ b/drivers/mmc/card/block.c |
| 52 | @@ -366,6 +366,22 @@ static int mmc_blk_issue_rw_rq(struct mm |
| 53 | if (brq.data.blocks > card->host->max_blk_count) |
| 54 | brq.data.blocks = card->host->max_blk_count; |
| 55 | |
| 56 | + if (mmc_host_is_spi(card->host)) { |
| 57 | + /* |
| 58 | + * Some SD-cards fail when we are reading last block |
| 59 | + * with multiblock read. In these cases we automatically |
| 60 | + * use single block reads. This only happens on SPI |
| 61 | + * hosts. |
| 62 | + */ |
| 63 | + if (rq_data_dir(req) == READ && brq.data.blocks > 1) { |
| 64 | + sector_t s = blk_rq_pos(req) + brq.data.blocks; |
| 65 | + |
| 66 | + if (s >= get_capacity(md->disk)) { |
| 67 | + disable_multi = 1; |
| 68 | + } |
| 69 | + } |
| 70 | + } |
| 71 | + |
| 72 | /* |
| 73 | * After a read error, we redo the request one sector at a time |
| 74 | * in order to accurately determine which sectors can be read |
| 75 | |
