[PATCH 2.6.18 2/2] mmc: Read mmc v4 EXT_CSD

[PATCH 2.6.18 2/2] mmc: Read mmc v4 EXT_CSD

[philipl]

Read in the mmc v4 EXT_CSD and populate the mmc_ext_csd.

--phil

Signed-off-by: Philip Langdale <philipl@overt.org>
---

 mmc.c |   84 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 84 insertions(+)

--- linux-old/drivers/mmc/mmc.c	2006-04-08 13:12:18.000000000 -0700 +++ linux/drivers/mmc/mmc.c	2006-04-08
13:12:25.000000000 -0700
@@ -953,6 +953,88 @@
 	}
 }

+static void mmc_read_ext_csds(struct mmc_host *host)
+{
+	int err;
+	struct mmc_card *card;
+
+	struct mmc_request mrq;
+	struct mmc_command cmd;
+	struct mmc_data data;
+
+	struct scatterlist sg;
+
+	/*
+	 * As the ext_csd is so large and mostly unused, we don't store the
+	 * raw block in mmc_card.
+	 */
+	u8 ext_csd[512];
+
+	list_for_each_entry(card, &host->cards, node) {
+		if (card->state & (MMC_STATE_DEAD|MMC_STATE_PRESENT))
+			continue;
+		if (card->csd.mmca_vsn < CSD_SPEC_VER_4)
+			continue;
+
+		err = mmc_select_card(host, card);
+		if (err != MMC_ERR_NONE) {
+			mmc_card_set_dead(card);
+			continue;
+		}
+
+		memset(&cmd, 0, sizeof(struct mmc_command));
+
+		cmd.opcode = MMC_SEND_EXT_CSD;
+		cmd.arg = 0;
+		cmd.flags = MMC_RSP_R1;
+
+		memset(&data, 0, sizeof(struct mmc_data));
+
+		data.timeout_ns = card->csd.tacc_ns * 10;
+		data.timeout_clks = card->csd.tacc_clks * 10;
+		data.blksz_bits = 9;
+		data.blocks = 1;
+		data.flags = MMC_DATA_READ;
+		data.sg = &sg;
+		data.sg_len = 1;
+
+		memset(&mrq, 0, sizeof(struct mmc_request));
+
+		mrq.cmd = &cmd;
+		mrq.data = &data;
+
+		sg_init_one(&sg, &ext_csd, 512);
+
+		mmc_wait_for_req(host, &mrq);
+
+		if (cmd.error != MMC_ERR_NONE || data.error != MMC_ERR_NONE) {
+			mmc_card_set_dead(card);
+			continue;
+		}
+
+		card->ext_csd.cmd_set_rev = ext_csd[EXT_CSD_CMD_SET_REV];
+		card->ext_csd.ext_csd_rev = ext_csd[EXT_CSD_EXT_CSD_REV];
+		card->ext_csd.csd_structure = ext_csd[EXT_CSD_CSD_STRUCTURE];
+		card->ext_csd.card_type = ext_csd[EXT_CSD_CARD_TYPE];
+
+		card->ext_csd.pwr_cl_52_195 = ext_csd[EXT_CSD_PWR_CL_52_195];
+		card->ext_csd.pwr_cl_26_195 = ext_csd[EXT_CSD_PWR_CL_26_195];
+		card->ext_csd.pwr_cl_52_360 = ext_csd[EXT_CSD_PWR_CL_52_360];
+		card->ext_csd.pwr_cl_26_360 = ext_csd[EXT_CSD_PWR_CL_26_360];
+
+		card->ext_csd.min_perf_r_4_26 = ext_csd[EXT_CSD_MIN_PERF_R_4_26];
+		card->ext_csd.min_perf_w_4_26 = ext_csd[EXT_CSD_MIN_PERF_W_4_26];
+		card->ext_csd.min_perf_r_8_26_4_52 = ext_csd[EXT_CSD_MIN_PERF_R_8_26_4_52];
+		card->ext_csd.min_perf_w_8_26_4_52 = ext_csd[EXT_CSD_MIN_PERF_W_8_26_4_52];
+		card->ext_csd.min_perf_r_8_52 = ext_csd[EXT_CSD_MIN_PERF_R_8_52];
+		card->ext_csd.min_perf_w_8_52 = ext_csd[EXT_CSD_MIN_PERF_W_8_52];
+
+		card->ext_csd.s_cmd_set = ext_csd[EXT_CSD_S_CMD_SET];
+	}
+
+	mmc_deselect_cards(host);
+}
+
 static void mmc_read_scrs(struct mmc_host *host)
 {
 	int err;
@@ -1153,6 +1420,8 @@

 	if (host->mode == MMC_MODE_SD)
 		mmc_read_scrs(host);
+	else
+		mmc_read_ext_csds(host);
 }

Re: [PATCH 2.6.18 2/2] mmc: Read mmc v4 EXT_CSD

[Pierre Ossman]

philipl@overt.org wrote:
> +	list_for_each_entry(card, &host->cards, node) {
> +		if (card->state & (MMC_STATE_DEAD|MMC_STATE_PRESENT))
> +			continue;
>   

Please use the macros.

> +		if (card->csd.mmca_vsn < CSD_SPEC_VER_4)
> +			continue;
> +
>   

You need to check that the card isn't SD before you can look at that
part of the csd structure. BUG_ON or similar is acceptable if you
consider it an error to call this function if SD cards are present.

> +		err = mmc_select_card(host, card);
> +		if (err != MMC_ERR_NONE) {
> +			mmc_card_set_dead(card);
> +			continue;
> +		}
> +
> +		memset(&cmd, 0, sizeof(struct mmc_command));
> +
> +		cmd.opcode = MMC_SEND_EXT_CSD;
> +		cmd.arg = 0;
> +		cmd.flags = MMC_RSP_R1;
> +
> +		memset(&data, 0, sizeof(struct mmc_data));
> +
> +		data.timeout_ns = card->csd.tacc_ns * 10;
> +		data.timeout_clks = card->csd.tacc_clks * 10;
>   

We have a new function for setting timeouts that you should use called
mmc_set_data_timeout().

Rgds
Pierre

Re: [PATCH 2.6.18 2/2] mmc: Read mmc v4 EXT_CSD

[Philip Langdale]

Pierre,

I will reorganise these patches into a series that detects and sets up
high-speed support. This is pretty simple. Wide-bus is harder because
you have to run their weird test pattern and check all the power class
stuff. Hopefully I'll have that done this week.

Pierre Ossman wrote:
> philipl@overt.org wrote:
>> +	list_for_each_entry(card, &host->cards, node) {
>> +		if (card->state & (MMC_STATE_DEAD|MMC_STATE_PRESENT))
>> +			continue;
>>   
> 
> Please use the macros.

I'm confused: mmc_read_scrs and mmc_read_csds use this exact same
construction in 2.6.18 and Linus' current tree. Should I be checking
another tree?

Will address all other comments.

Thanks,

--phil

Re: [PATCH 2.6.18 2/2] mmc: Read mmc v4 EXT_CSD

[Pierre Ossman]

Philip Langdale wrote:
> Pierre,
>
> I will reorganise these patches into a series that detects and sets up
> high-speed support. This is pretty simple. Wide-bus is harder because
> you have to run their weird test pattern and check all the power class
> stuff. Hopefully I'll have that done this week.
>   

Goodie. Have these patches seen any wide spread testing?

> Pierre Ossman wrote:
>   
>> philipl@overt.org wrote:
>>     
>>> +	list_for_each_entry(card, &host->cards, node) {
>>> +		if (card->state & (MMC_STATE_DEAD|MMC_STATE_PRESENT))
>>> +			continue;
>>>   
>>>       
>> Please use the macros.
>>     
>
> I'm confused: mmc_read_scrs and mmc_read_csds use this exact same
> construction in 2.6.18 and Linus' current tree. Should I be checking
> another tree?
>
>   

Nah, they're also wrong. Patches welcome ;)

Rgds
Pierre

Re: [PATCH 2.6.18 2/2] mmc: Read mmc v4 EXT_CSD

[Frantisek Dufka]

Pierre Ossman wrote:
 >
 >Goodie. Have these patches seen any wide spread testing?

Several people tried in on Nokia N770. So far there were no error 
reports and some successful ones. Speed selection (52 vs 26Mhz) seems to 
work fine.

Succesful report examples:
http://www.gossamer-threads.com/lists/maemo/developers/11613#11613
http://www.gossamer-threads.com/lists/maemo/developers/11573#11573

I wouldn't call it widespread testing but it definitely works for some 
people.

Frantisek