Skip to contentSkip to conversation

linux-next: rsi: fix using of removed stuff from mmc

PATCH In review Seungwon Jeon 2 participants 6 messages Apr 22, 2014

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 linux-next: rsi: fix using of removed stuff from mmc

As commit 3957848(mmc: drop the speed mode of card's state) applies, this change should be followed.

drivers/net/wireless/rsi/rsi_91x_sdio.c:288:20: error: 'MMC_STATE_HIGHSPEED' undeclared (first use in this function) drivers/net/wireless/rsi/rsi_91x_sdio.c:299:4: error: implicit declaration of function 'mmc_card_set_highspeed' [-Werror=implicit-function-declaration] drivers/net/wireless/rsi/rsi_91x_sdio.c:306:2: error: implicit declaration of function 'mmc_card_highspeed' [-Werror=implicit-function-declaration]

---
 drivers/net/wireless/rsi/rsi_91x_sdio.c |    4 +---
 1 files changed, 1 insertions(+), 3 deletions(-)
drivers/net/wireless/rsi/rsi_91x_sdio.c +1 -3
index 2e39d38..46e7af4 100644
@@ -285,7 +285,6 @@ static void rsi_reset_card(struct sdio_func *pfunction)
 		if (err) {
 			rsi_dbg(ERR_ZONE, "%s: CCCR speed reg read failed: %d\n",
 				__func__, err);
-			card->state &= ~MMC_STATE_HIGHSPEED;
 		} else {
 			err = rsi_cmd52writebyte(card,
 						 SDIO_CCCR_SPEED,
@@ -296,14 +295,13 @@ static void rsi_reset_card(struct sdio_func *pfunction)
 					__func__, err);
 				return;
 			}
-			mmc_card_set_highspeed(card);
 			host->ios.timing = MMC_TIMING_SD_HS;
 			host->ops->set_ios(host, &host->ios);
 		}
 	}
 
 	/* Set clock */
-	if (mmc_card_highspeed(card))
+	if (mmc_card_hs(card))
 		clock = 50000000;
 	else
 		clock = card->cis.max_dtr;
Signature · 1 line
1.7.0.4
#1 lore raw
Quoted 8 lines · Seungwon Jeon

As commit 3957848(mmc: drop the speed mode of card's state) applies, this change should be followed.

drivers/net/wireless/rsi/rsi_91x_sdio.c:288:20: error: 'MMC_STATE_HIGHSPEED' undeclared (first use in this function) drivers/net/wireless/rsi/rsi_91x_sdio.c:299:4: error: implicit declaration of function 'mmc_card_set_highspeed' [-Werror=implicit-function-declaration] drivers/net/wireless/rsi/rsi_91x_sdio.c:306:2: error: implicit declaration of function 'mmc_card_highspeed' [-Werror=implicit-function-declaration]

Hi Seungwon,

Thanks for responding quickly!

Quoted 33 lines
---
 drivers/net/wireless/rsi/rsi_91x_sdio.c |    4 +---
 1 files changed, 1 insertions(+), 3 deletions(-)
drivers/net/wireless/rsi/rsi_91x_sdio.c +1 -4
index 2e39d38..46e7af4 100644
@@ -285,7 +285,6 @@ static void rsi_reset_card(struct sdio_func *pfunction)
                if (err) {
                        rsi_dbg(ERR_ZONE, "%s: CCCR speed reg read failed: %d\n",
                                __func__, err);
-                       card->state &= ~MMC_STATE_HIGHSPEED;
                } else {
                        err = rsi_cmd52writebyte(card,
                                                 SDIO_CCCR_SPEED,
@@ -296,14 +295,13 @@ static void rsi_reset_card(struct sdio_func *pfunction)
                                        __func__, err);
                                return;
                        }
-                       mmc_card_set_highspeed(card);
                        host->ios.timing = MMC_TIMING_SD_HS;
                        host->ops->set_ios(host, &host->ios);
                }
        }
 
        /* Set clock */
-       if (mmc_card_highspeed(card))
+       if (mmc_card_hs(card))
                clock = 50000000;
        else
                clock = card->cis.max_dtr;
--
Quoted 3 lines

1.7.0.4

I am not sure it's safe to carry this patch through Chris' mmc tree due to merge conflicts with John's wireless tree.

Though, since you have changed the mmc_card_highspeed() function to be named to mmc_card_hs(), we need to work out the dependency.

We have some options to handle this, I suggest the following.

Re-spin this patch to keep using mmc_card_highspeed() and let John take it through his wireless tree. Thus you also need to keep the name mmc_card_highspeed() function from the patches to the mmc core - could you please re-spin and post new version of those patches as well!?

Kind regards
Ulf Hansson

#2 lore raw
Quoted 57 lines · Ulf Hansson
Quoted 4 lines · Seungwon Jeon

As commit 3957848(mmc: drop the speed mode of card's state) applies, this change should be followed.

drivers/net/wireless/rsi/rsi_91x_sdio.c:288:20: error: 'MMC_STATE_HIGHSPEED' undeclared (first use

in this function)

Quoted 1 line

drivers/net/wireless/rsi/rsi_91x_sdio.c:299:4: error: implicit declaration of function

'mmc_card_set_highspeed' [-Werror=implicit-function-declaration]

Quoted 1 line

drivers/net/wireless/rsi/rsi_91x_sdio.c:306:2: error: implicit declaration of function

'mmc_card_highspeed' [-Werror=implicit-function-declaration]

Quoted 2 lines

Hi Seungwon,

Thanks for responding quickly!

Quoted 33 lines
---
 drivers/net/wireless/rsi/rsi_91x_sdio.c |    4 +---
 1 files changed, 1 insertions(+), 3 deletions(-)
drivers/net/wireless/rsi/rsi_91x_sdio.c +1 -4
index 2e39d38..46e7af4 100644
@@ -285,7 +285,6 @@ static void rsi_reset_card(struct sdio_func *pfunction)
                if (err) {
                        rsi_dbg(ERR_ZONE, "%s: CCCR speed reg read failed: %d\n",
                                __func__, err);
-                       card->state &= ~MMC_STATE_HIGHSPEED;
                } else {
                        err = rsi_cmd52writebyte(card,
                                                 SDIO_CCCR_SPEED,
@@ -296,14 +295,13 @@ static void rsi_reset_card(struct sdio_func *pfunction)
                                        __func__, err);
                                return;
                        }
-                       mmc_card_set_highspeed(card);
                        host->ios.timing = MMC_TIMING_SD_HS;
                        host->ops->set_ios(host, &host->ios);
                }
        }
 
        /* Set clock */
-       if (mmc_card_highspeed(card))
+       if (mmc_card_hs(card))
                clock = 50000000;
        else
                clock = card->cis.max_dtr;
--
Quoted 3 lines

1.7.0.4

I am not sure it's safe to carry this patch through Chris' mmc tree due to merge conflicts with John's wireless tree.

Quoted 7 lines

Though, since you have changed the mmc_card_highspeed() function to be named to mmc_card_hs(), we need to work out the dependency.

We have some options to handle this, I suggest the following.

Re-spin this patch to keep using mmc_card_highspeed() and let John take it through his wireless tree. Thus you also need to keep the

It would be better if this patch can be handled in Chris's.
Because if the rest of changes of this patch with keeping "mmc_card_highspeed" is applied, high-speed SDIO will be ignored in John's tree.

Thanks,
Seungwon Jeon

Quoted 6 lines

name mmc_card_highspeed() function from the patches to the mmc core - could you please re-spin and post new version of those patches as well!?

Kind regards
Ulf Hansson

#3 lore raw
Quoted 61 lines · Seungwon Jeon
Quoted 60 lines · Ulf Hansson
Quoted 4 lines · Seungwon Jeon

As commit 3957848(mmc: drop the speed mode of card's state) applies, this change should be followed.

drivers/net/wireless/rsi/rsi_91x_sdio.c:288:20: error: 'MMC_STATE_HIGHSPEED' undeclared (first use

in this function)

Quoted 1 line

drivers/net/wireless/rsi/rsi_91x_sdio.c:299:4: error: implicit declaration of function

'mmc_card_set_highspeed' [-Werror=implicit-function-declaration]

Quoted 1 line

drivers/net/wireless/rsi/rsi_91x_sdio.c:306:2: error: implicit declaration of function

'mmc_card_highspeed' [-Werror=implicit-function-declaration]

Quoted 2 lines

Hi Seungwon,

Thanks for responding quickly!

Quoted 33 lines
---
 drivers/net/wireless/rsi/rsi_91x_sdio.c |    4 +---
 1 files changed, 1 insertions(+), 3 deletions(-)
drivers/net/wireless/rsi/rsi_91x_sdio.c +1 -4
index 2e39d38..46e7af4 100644
@@ -285,7 +285,6 @@ static void rsi_reset_card(struct sdio_func *pfunction)
                if (err) {
                        rsi_dbg(ERR_ZONE, "%s: CCCR speed reg read failed: %d\n",
                                __func__, err);
-                       card->state &= ~MMC_STATE_HIGHSPEED;
                } else {
                        err = rsi_cmd52writebyte(card,
                                                 SDIO_CCCR_SPEED,
@@ -296,14 +295,13 @@ static void rsi_reset_card(struct sdio_func *pfunction)
                                        __func__, err);
                                return;
                        }
-                       mmc_card_set_highspeed(card);
                        host->ios.timing = MMC_TIMING_SD_HS;
                        host->ops->set_ios(host, &host->ios);
                }
        }
 
        /* Set clock */
-       if (mmc_card_highspeed(card))
+       if (mmc_card_hs(card))
                clock = 50000000;
        else
                clock = card->cis.max_dtr;
--
Quoted 3 lines

1.7.0.4

I am not sure it's safe to carry this patch through Chris' mmc tree due to merge conflicts with John's wireless tree.

Though, since you have changed the mmc_card_highspeed() function to be named to mmc_card_hs(), we need to work out the dependency.

Quoted 7 lines
Quoted 4 lines

We have some options to handle this, I suggest the following.

Re-spin this patch to keep using mmc_card_highspeed() and let John take it through his wireless tree. Thus you also need to keep the

It would be better if this patch can be handled in Chris's.
Because if the rest of changes of this patch with keeping "mmc_card_highspeed" is applied, high-speed SDIO will be ignored in John's tree.

You are right Seungwon!

Please re-spin the patchset on the mmc core an squash the code from this patch, into the patch "mmc: drop the speed mode of card's state". We then need to get acks from John and Fariya.

BTW, I had a look at the rsi_reset_card() function (drivers/net/wireless/rsi/rsi_91x_sdio.c), which handles a complete SDIO re-initialization and I think performs "layering violations" while doing that. I suppose it's because of lack of documentation about the mmc/sdio core, but I really don't think an SDIO func driver should be doing that kind of stuff by itself. There are APIs to use to perform an SDIO reset. Two options exists:

a) Use pm_runtime_get|put in combination with a MMC_CAP_POWER_OFF_CARD enabled host.
b) Invoke mmc_power_save|restore_host() API.

Kind regards
Ulf Hansson

Quoted 11 lines

Thanks,
Seungwon Jeon

Quoted 6 lines

name mmc_card_highspeed() function from the patches to the mmc core - could you please re-spin and post new version of those patches as well!?

Kind regards
Ulf Hansson

#4 lore raw
Quoted 71 lines · Ulf Hansson
Quoted 6 lines · Seungwon Jeon
Quoted 5 lines · Ulf Hansson
Quoted 4 lines · Seungwon Jeon

As commit 3957848(mmc: drop the speed mode of card's state) applies, this change should be followed.

drivers/net/wireless/rsi/rsi_91x_sdio.c:288:20: error: 'MMC_STATE_HIGHSPEED' undeclared (first

use

Quoted 55 lines
Quoted 55 lines

in this function)

Quoted 1 line

drivers/net/wireless/rsi/rsi_91x_sdio.c:299:4: error: implicit declaration of function

'mmc_card_set_highspeed' [-Werror=implicit-function-declaration]

Quoted 1 line

drivers/net/wireless/rsi/rsi_91x_sdio.c:306:2: error: implicit declaration of function

'mmc_card_highspeed' [-Werror=implicit-function-declaration]

Quoted 2 lines

Hi Seungwon,

Thanks for responding quickly!

Quoted 33 lines
---
 drivers/net/wireless/rsi/rsi_91x_sdio.c |    4 +---
 1 files changed, 1 insertions(+), 3 deletions(-)
drivers/net/wireless/rsi/rsi_91x_sdio.c +1 -4
index 2e39d38..46e7af4 100644
@@ -285,7 +285,6 @@ static void rsi_reset_card(struct sdio_func *pfunction)
                if (err) {
                        rsi_dbg(ERR_ZONE, "%s: CCCR speed reg read failed: %d\n",
                                __func__, err);
-                       card->state &= ~MMC_STATE_HIGHSPEED;
                } else {
                        err = rsi_cmd52writebyte(card,
                                                 SDIO_CCCR_SPEED,
@@ -296,14 +295,13 @@ static void rsi_reset_card(struct sdio_func *pfunction)
                                        __func__, err);
                                return;
                        }
-                       mmc_card_set_highspeed(card);
                        host->ios.timing = MMC_TIMING_SD_HS;
                        host->ops->set_ios(host, &host->ios);
                }
        }
 
        /* Set clock */
-       if (mmc_card_highspeed(card))
+       if (mmc_card_hs(card))
                clock = 50000000;
        else
                clock = card->cis.max_dtr;
--
Quoted 3 lines

1.7.0.4

I am not sure it's safe to carry this patch through Chris' mmc tree due to merge conflicts with John's wireless tree.

Though, since you have changed the mmc_card_highspeed() function to be named to mmc_card_hs(), we need to work out the dependency.

Quoted 7 lines
Quoted 4 lines

We have some options to handle this, I suggest the following.

Re-spin this patch to keep using mmc_card_highspeed() and let John take it through his wireless tree. Thus you also need to keep the

It would be better if this patch can be handled in Chris's.
Because if the rest of changes of this patch with keeping "mmc_card_highspeed" is applied, high-speed SDIO will be ignored in John's tree.

Quoted 5 lines

You are right Seungwon!

Please re-spin the patchset on the mmc core an squash the code from this patch, into the patch "mmc: drop the speed mode of card's state". We then need to get acks from John and Fariya.

Sure! I'll send before long.
You still want to get back naming for mmc_card_highspeed?

Quoted 9 lines

BTW, I had a look at the rsi_reset_card() function (drivers/net/wireless/rsi/rsi_91x_sdio.c), which handles a complete SDIO re-initialization and I think performs "layering violations" while doing that. I suppose it's because of lack of documentation about the mmc/sdio core, but I really don't think an SDIO func driver should be doing that kind of stuff by itself. There are APIs to use to perform an SDIO reset. Two options exists:

Quoted 3 lines

a) Use pm_runtime_get|put in combination with a MMC_CAP_POWER_OFF_CARD enabled host.
b) Invoke mmc_power_save|restore_host() API.

Yes, we could expect better changes in rsi sdio soon.

Thanks,
Seungwon Jeon

Quoted 15 lines

Kind regards
Ulf Hansson

Quoted 11 lines

Thanks,
Seungwon Jeon

Quoted 6 lines

name mmc_card_highspeed() function from the patches to the mmc core - could you please re-spin and post new version of those patches as well!?

Kind regards
Ulf Hansson

#5 lore raw
Quoted 15 lines
Quoted 15 lines
Quoted 14 lines
Quoted 11 lines

I am not sure it's safe to carry this patch through Chris' mmc tree due to merge conflicts with John's wireless tree.

Though, since you have changed the mmc_card_highspeed() function to be named to mmc_card_hs(), we need to work out the dependency.

We have some options to handle this, I suggest the following.

Re-spin this patch to keep using mmc_card_highspeed() and let John take it through his wireless tree. Thus you also need to keep the

It would be better if this patch can be handled in Chris's.
Because if the rest of changes of this patch with keeping "mmc_card_highspeed" is applied, high-speed SDIO will be ignored in John's tree.

Quoted 7 lines
Quoted 5 lines

You are right Seungwon!

Please re-spin the patchset on the mmc core an squash the code from this patch, into the patch "mmc: drop the speed mode of card's state". We then need to get acks from John and Fariya.

Sure! I'll send before long.
You still want to get back naming for mmc_card_highspeed?

It's up to you, I have no strong opinion.

Kind regards
Ulf Hansson

Quoted 33 lines
Quoted 10 lines

BTW, I had a look at the rsi_reset_card() function (drivers/net/wireless/rsi/rsi_91x_sdio.c), which handles a complete SDIO re-initialization and I think performs "layering violations" while doing that. I suppose it's because of lack of documentation about the mmc/sdio core, but I really don't think an SDIO func driver should be doing that kind of stuff by itself. There are APIs to use to perform an SDIO reset. Two options exists:

Quoted 3 lines

a) Use pm_runtime_get|put in combination with a MMC_CAP_POWER_OFF_CARD enabled host.
b) Invoke mmc_power_save|restore_host() API.

Yes, we could expect better changes in rsi sdio soon.

Thanks,
Seungwon Jeon

Quoted 15 lines

Kind regards
Ulf Hansson

Quoted 11 lines

Thanks,
Seungwon Jeon

Quoted 6 lines

name mmc_card_highspeed() function from the patches to the mmc core - could you please re-spin and post new version of those patches as well!?

Kind regards
Ulf Hansson