Skip to contentSkip to conversation

Fix operation on Odroid devices

PATCH Superseded Tobias Jakobi 5 participants 14 messages Sep 20, 2015 – Sep 24, 2015 MMC, SD and SDIO

v1 of 2 Superseded Latest is v2 →

lore
Share & tools
lore
#0 lore raw

cover Fix operation on Odroid devices

Hello,

currently operation on Exynos4412-based Odroid devices is broken.

The bootloader stops with this message:
Card did not respond to voltage select!
*** Warning - MMC init failed, using default environment

This series fixes error handling in the s5p sdhci driver and the cause of the issue, which is an inverted card detection check.

Thanks goes to Marek Vasut and Sjoerd Simons who helped me on IRC to get this triaged.

With best wishes,
Tobias

Tobias Jakobi (4):
exynos: Properly initialize host_caps in s5p_sdhci_core_init()
exynos: Fix passing of errors in exynos_mmc_init()
exynos: be more verbose in process_nodes()
exynos: fix and cleanup do_sdhci_init()

 drivers/mmc/s5p_sdhci.c | 39 +++++++++++++++++++++++----------------
 1 file changed, 23 insertions(+), 16 deletions(-)
Signature · 1 line
2.0.5
#1 lore raw

PATCH 1/4 exynos: Properly initialize host_caps in s5p_sdhci_core_init()

The sdhci_host struct is allocated in s5p_sdhci_init() but the fields are not initialized.

---
 drivers/mmc/s5p_sdhci.c | 1 +
 1 file changed, 1 insertion(+)
drivers/mmc/s5p_sdhci.c +1 -0
index 4db51d6..e9c43a9 100644
@@ -76,6 +76,7 @@ static int s5p_sdhci_core_init(struct sdhci_host *host)
 	host->set_control_reg = &s5p_sdhci_set_control_reg;
 	host->set_clock = set_mmc_clk;
 
+	host->host_caps = 0;
 	if (host->bus_width == 8)
 		host->host_caps |= MMC_MODE_8BIT;
 
Signature · 1 line
2.0.5
#2 lore raw

PATCH 1/4 exynos: Properly initialize host_caps in s5p_sdhci_core_init()

Hi Tobias,

Quoted 13 lines

The sdhci_host struct is allocated in s5p_sdhci_init() but the fields are not initialized.

---
 drivers/mmc/s5p_sdhci.c | 1 +
 1 file changed, 1 insertion(+)
drivers/mmc/s5p_sdhci.c +0 -0
index 4db51d6..e9c43a9 100644

drivers/mmc/s5p_sdhci.c static int s5p_sdhci_core_init(struct sdhci_host · line 82

Quoted 7 lines
drivers/mmc/s5p_sdhci.c +1 -0
@@ -76,6 +76,7 @@ static int s5p_sdhci_core_init(struct sdhci_host
 *host) host->set_control_reg = &s5p_sdhci_set_control_reg;
 	host->set_clock = set_mmc_clk;
 
+	host->host_caps = 0;
 	if (host->bus_width == 8)
 		host->host_caps |= MMC_MODE_8BIT;
 
Signature · 5 lines
Best regards,

Lukasz Majewski

Samsung R&D Institute Poland (SRPOL) | Linux Platform Group
#3 lore raw

PATCH 2/4 exynos: Fix passing of errors in exynos_mmc_init()

In reply to Tobias Jakobi

exynos_mmc_init() always returns zero, so for the caller it looks like it never fails.

Correct this by returning the error code of process_nodes(). For process_nodes() do something similar and return early when do_sdhci_init() fails.

---
 drivers/mmc/s5p_sdhci.c | 10 ++++++----
 1 file changed, 6 insertions(+), 4 deletions(-)
drivers/mmc/s5p_sdhci.c +6 -4
index e9c43a9..bc2102a 100644
@@ -187,7 +187,11 @@ static int process_nodes(const void *blob, int node_list[], int count)
 			printf("%s: failed to decode dev %d\n",	__func__, i);
 			return -1;
 		}
-		do_sdhci_init(host);
+
+		if (do_sdhci_init(host)) {
+			printf("%s: failed to initialize dev %d\n", __func__, i);
+			return -2;
+		}
 	}
 	return 0;
 }
@@ -201,8 +205,6 @@ int exynos_mmc_init(const void *blob)
 			COMPAT_SAMSUNG_EXYNOS_MMC, node_list,
 			SDHCI_MAX_HOSTS);
 
-	process_nodes(blob, node_list, count);
-
-	return 0;
+	return process_nodes(blob, node_list, count);
 }
 #endif
Signature · 1 line
2.0.5
#4 lore raw

PATCH 2/4 exynos: Fix passing of errors in exynos_mmc_init()

Hi Tobias,

Quoted 25 lines

exynos_mmc_init() always returns zero, so for the caller it looks like it never fails.

Correct this by returning the error code of process_nodes(). For process_nodes() do something similar and return early when do_sdhci_init() fails.

---
 drivers/mmc/s5p_sdhci.c | 10 ++++++----
 1 file changed, 6 insertions(+), 4 deletions(-)
drivers/mmc/s5p_sdhci.c +0 -0
index e9c43a9..bc2102a 100644
@@ -187,7 +187,11 @@ static int process_nodes(const void *blob, int

node_list[], int count) printf("%s: failed to decode dev
%d\n", __func__, i); return -1;

drivers/mmc/s5p_sdhci.c +3 -1
 		}
-		do_sdhci_init(host);
+
+		if (do_sdhci_init(host)) {
+			printf("%s: failed to initialize dev %d\n",

__func__, i);

drivers/mmc/s5p_sdhci.c static int process_nodes(const void *blob, int node_list[], int count) · line 193

Quoted 1 line
drivers/mmc/s5p_sdhci.c +1 -0
+			return -2;

IMHO, it would be better to write this code as follows:

ret = do_sdhci_init(host);
if (ret) {

	printf();
	return ret;
}

In the above code you read the exact return code from do_sdhci_init() and then you pass it to upper layer.

Returning only -2 is far less informational.

Quoted 14 lines
drivers/mmc/s5p_sdhci.c +2 -3
+		}
 	}
 	return 0;
 }
@@ -201,8 +205,6 @@ int exynos_mmc_init(const void *blob)
 			COMPAT_SAMSUNG_EXYNOS_MMC, node_list,
 			SDHCI_MAX_HOSTS);
 
-	process_nodes(blob, node_list, count);
-
-	return 0;
+	return process_nodes(blob, node_list, count);
 }
 #endif
Signature · 5 lines
Best regards,

Lukasz Majewski

Samsung R&D Institute Poland (SRPOL) | Linux Platform Group
#5 lore raw

PATCH 2/4 exynos: Fix passing of errors in exynos_mmc_init()

Hi.

Quoted 31 lines · Lukasz Majewski

Hi Tobias,

Quoted 25 lines

exynos_mmc_init() always returns zero, so for the caller it looks like it never fails.

Correct this by returning the error code of process_nodes(). For process_nodes() do something similar and return early when do_sdhci_init() fails.

---
 drivers/mmc/s5p_sdhci.c | 10 ++++++----
 1 file changed, 6 insertions(+), 4 deletions(-)
drivers/mmc/s5p_sdhci.c +0 -0
index e9c43a9..bc2102a 100644
@@ -187,7 +187,11 @@ static int process_nodes(const void *blob, int

node_list[], int count) printf("%s: failed to decode dev
%d\n", __func__, i); return -1;

drivers/mmc/s5p_sdhci.c +3 -1
 		}
-		do_sdhci_init(host);
+
+		if (do_sdhci_init(host)) {
+			printf("%s: failed to initialize dev %d\n",

__func__, i);

drivers/mmc/s5p_sdhci.c static int process_nodes(const void *blob, int node_list[], int count) · line 193

Quoted 1 line
drivers/mmc/s5p_sdhci.c +1 -0
+			return -2;

IMHO, it would be better to write this code as follows:

Quoted 6 lines

ret = do_sdhci_init(host);
if (ret) {

	printf();
	return ret;
}

I think it should be replaced to "continue;", not "return ret;"
If returned the fail, then next host can't initialize.(if there is next host..) So maybe, it didn't use "return ret".

Best Regards,
Jaehoon Chung

Quoted 23 lines

In the above code you read the exact return code from do_sdhci_init() and then you pass it to upper layer.

Returning only -2 is far less informational.

Quoted 14 lines
drivers/mmc/s5p_sdhci.c +2 -3
+		}
 	}
 	return 0;
 }
@@ -201,8 +205,6 @@ int exynos_mmc_init(const void *blob)
 			COMPAT_SAMSUNG_EXYNOS_MMC, node_list,
 			SDHCI_MAX_HOSTS);
 
-	process_nodes(blob, node_list, count);
-
-	return 0;
+	return process_nodes(blob, node_list, count);
 }
 #endif
#6 lore raw

PATCH 2/4 exynos: Fix passing of errors in exynos_mmc_init()

Quoted 37 lines · Jaehoon Chung

Hi.

Quoted 34 lines · Lukasz Majewski

Hi Tobias,

Quoted 25 lines

exynos_mmc_init() always returns zero, so for the caller it looks like it never fails.

Correct this by returning the error code of process_nodes(). For process_nodes() do something similar and return early when do_sdhci_init() fails.

---
 drivers/mmc/s5p_sdhci.c | 10 ++++++----
 1 file changed, 6 insertions(+), 4 deletions(-)
drivers/mmc/s5p_sdhci.c +0 -0
index e9c43a9..bc2102a 100644
@@ -187,7 +187,11 @@ static int process_nodes(const void *blob, int

node_list[], int count) printf("%s: failed to decode dev
%d\n", __func__, i); return -1;

drivers/mmc/s5p_sdhci.c +3 -1
 		}
-		do_sdhci_init(host);
+
+		if (do_sdhci_init(host)) {
+			printf("%s: failed to initialize dev %d\n",

__func__, i);

drivers/mmc/s5p_sdhci.c static int process_nodes(const void *blob, int node_list[], int count) · line 193

Quoted 1 line
drivers/mmc/s5p_sdhci.c +1 -0
+			return -2;

IMHO, it would be better to write this code as follows:

ret = do_sdhci_init(host);
if (ret) {

Quoted 8 lines
Quoted 3 lines
	printf();
	return ret;
}

I think it should be replaced to "continue;", not "return ret;"
If returned the fail, then next host can't initialize.(if there is next host..) So maybe, it didn't use "return ret".

Right.
If succeeded to initialize at least one host, should not be returned an error.

Thanks,
Minkyu Kang.

#7 lore raw

cover Fix operation on Odroid devices

In reply to Tobias Jakobi

Hi,

Quoted 7 lines · Tobias Jakobi

Hello,

currently operation on Exynos4412-based Odroid devices is broken.

The bootloader stops with this message:
Card did not respond to voltage select!
*** Warning - MMC init failed, using default environment

Which board did you use? and Which cad is failed? SD or eMMC? Your [PATCH 4/4] is strange..so i want to get more information.

If you can share more information, i can explain more exactly about your problem.

Best Regards,
Jaehoon Chung

Quoted 20 lines

This series fixes error handling in the s5p sdhci driver and the cause of the issue, which is an inverted card detection check.

Thanks goes to Marek Vasut and Sjoerd Simons who helped me on IRC to get this triaged.

With best wishes,
Tobias

Tobias Jakobi (4):
exynos: Properly initialize host_caps in s5p_sdhci_core_init()
exynos: Fix passing of errors in exynos_mmc_init()
exynos: be more verbose in process_nodes()
exynos: fix and cleanup do_sdhci_init()

 drivers/mmc/s5p_sdhci.c | 39 +++++++++++++++++++++++----------------
 1 file changed, 23 insertions(+), 16 deletions(-)
#8 lore raw

cover Fix operation on Odroid devices

Hello,

Quoted 5 lines · Jaehoon Chung

Hi,

Quoted 2 lines · Tobias Jakobi

Hello,

Quoted 7 lines
Quoted 5 lines

currently operation on Exynos4412-based Odroid devices is broken.

The bootloader stops with this message:
Card did not respond to voltage select!
*** Warning - MMC init failed, using default environment

Which board did you use? and Which cad is failed? SD or eMMC?

This is an Odroid-X2. I'm booting from SD with a Sandisk Extreme. I don't have any eMMC here.

Actually it was some other person who encountered this issue with his Odroid-X2 and booting from SD. After he made me aware of it, I also updated my u-boot (I was then using a version from May 2015, which works perfectly btw) and encountered the very same issue as him.

Quoted 1 line

Your [PATCH 4/4] is strange..so i want to get more information.

For me it restores proper operation of the bootloader to the board.

With best wishes,
Tobias

Quoted 27 lines

If you can share more information, i can explain more exactly about your problem.

Best Regards,
Jaehoon Chung

Quoted 20 lines

This series fixes error handling in the s5p sdhci driver and the cause of the issue, which is an inverted card detection check.

Thanks goes to Marek Vasut and Sjoerd Simons who helped me on IRC to get this triaged.

With best wishes,
Tobias

Tobias Jakobi (4):
exynos: Properly initialize host_caps in s5p_sdhci_core_init()
exynos: Fix passing of errors in exynos_mmc_init()
exynos: be more verbose in process_nodes()
exynos: fix and cleanup do_sdhci_init()

 drivers/mmc/s5p_sdhci.c | 39 +++++++++++++++++++++++----------------
 1 file changed, 23 insertions(+), 16 deletions(-)
#9 lore raw

cover Fix operation on Odroid devices

Dear, Tobias.

Quoted 10 lines · Tobias Jakobi

Hello,

Quoted 7 lines · Jaehoon Chung

Hi,

Quoted 4 lines · Tobias Jakobi

Hello,

currently operation on Exynos4412-based Odroid devices is broken.

Quoted 7 lines
Quoted 5 lines
Quoted 3 lines

The bootloader stops with this message:
Card did not respond to voltage select!
*** Warning - MMC init failed, using default environment

Which board did you use? and Which cad is failed? SD or eMMC?

This is an Odroid-X2. I'm booting from SD with a Sandisk Extreme. I don't have any eMMC here.

But if your patch is applied, other exynos4412 and exynos using s5p_sdhci don't work fine for SD-card. This case is the specific X2 board. I think schematic is wrong.
(I have checked the Odrodd_X-base_Rev_06.pdf)

Commonly, SD-detect pin is supplied to always-on power.
But odroid-x/u board used the same supplier with I/O line. It's wrong.

This is H/W mis-designed problem, not code problem.
Well, I will reproduce your problem with X2 board.
And if reproduced your problem, I will find other solution, not this. how about?

Quoted 6 lines

Actually it was some other person who encountered this issue with his Odroid-X2 and booting from SD. After he made me aware of it, I also updated my u-boot (I was then using a version from May 2015, which works perfectly btw) and encountered the very same issue as him.

Need to consider other SoC. You seems to consider only Odroid based-on exyno4412. Did you test other exynos4412 board?

Best Regards,
Jaehoon Chung

Quoted 39 lines
Quoted 3 lines

Your [PATCH 4/4] is strange..so i want to get more information.

For me it restores proper operation of the bootloader to the board.

With best wishes,
Tobias

Quoted 27 lines

If you can share more information, i can explain more exactly about your problem.

Best Regards,
Jaehoon Chung

Quoted 20 lines

This series fixes error handling in the s5p sdhci driver and the cause of the issue, which is an inverted card detection check.

Thanks goes to Marek Vasut and Sjoerd Simons who helped me on IRC to get this triaged.

With best wishes,
Tobias

Tobias Jakobi (4):
exynos: Properly initialize host_caps in s5p_sdhci_core_init()
exynos: Fix passing of errors in exynos_mmc_init()
exynos: be more verbose in process_nodes()
exynos: fix and cleanup do_sdhci_init()

 drivers/mmc/s5p_sdhci.c | 39 +++++++++++++++++++++++----------------
 1 file changed, 23 insertions(+), 16 deletions(-)
#10 lore raw

cover Fix operation on Odroid devices

Hello Jaehoon,

Quoted 21 lines · Jaehoon Chung

Dear, Tobias.

Quoted 10 lines · Tobias Jakobi

Hello,

Quoted 7 lines · Jaehoon Chung

Hi,

Quoted 4 lines · Tobias Jakobi

Hello,

currently operation on Exynos4412-based Odroid devices is broken.

Quoted 7 lines
Quoted 5 lines
Quoted 3 lines

The bootloader stops with this message:
Card did not respond to voltage select!
*** Warning - MMC init failed, using default environment

Which board did you use? and Which cad is failed? SD or eMMC?

This is an Odroid-X2. I'm booting from SD with a Sandisk Extreme. I don't have any eMMC here.

Quoted 6 lines

But if your patch is applied, other exynos4412 and exynos using s5p_sdhci don't work fine for SD-card. This case is the specific X2 board. I think schematic is wrong.
(I have checked the Odrodd_X-base_Rev_06.pdf)

Commonly, SD-detect pin is supplied to always-on power.
But odroid-x/u board used the same supplier with I/O line. It's wrong.

I wouldn't know, I don't even know how to properly read these schematics.

Quoted 3 lines

This is H/W mis-designed problem, not code problem.
Well, I will reproduce your problem with X2 board.
And if reproduced your problem, I will find other solution, not this. how about?

Sure, as long as this fixes the issue. Let me know if I can test anything.

Quoted 8 lines
Quoted 5 lines

Actually it was some other person who encountered this issue with his Odroid-X2 and booting from SD. After he made me aware of it, I also updated my u-boot (I was then using a version from May 2015, which works perfectly btw) and encountered the very same issue as him.

Need to consider other SoC. You seems to consider only Odroid based-on exyno4412. Did you test other exynos4412 board?

I have only one ARM board here, which is the Odroid-X2. Nothing else.

With best wishes,
Tobias

Quoted 44 lines

Best Regards,
Jaehoon Chung

Quoted 39 lines
Quoted 3 lines

Your [PATCH 4/4] is strange..so i want to get more information.

For me it restores proper operation of the bootloader to the board.

With best wishes,
Tobias

Quoted 27 lines

If you can share more information, i can explain more exactly about your problem.

Best Regards,
Jaehoon Chung

Quoted 20 lines

This series fixes error handling in the s5p sdhci driver and the cause of the issue, which is an inverted card detection check.

Thanks goes to Marek Vasut and Sjoerd Simons who helped me on IRC to get this triaged.

With best wishes,
Tobias

Tobias Jakobi (4):
exynos: Properly initialize host_caps in s5p_sdhci_core_init()
exynos: Fix passing of errors in exynos_mmc_init()
exynos: be more verbose in process_nodes()
exynos: fix and cleanup do_sdhci_init()

 drivers/mmc/s5p_sdhci.c | 39 +++++++++++++++++++++++----------------
 1 file changed, 23 insertions(+), 16 deletions(-)
#12 lore raw

cover Fix operation on Odroid devices

Hi, Tobias,

Quoted 1 line · Tobias Jakobi

Forgot to ask this: What about the first three patches?

Will check and review them.

Best Regards,
Jaehoon Chung

Quoted 4 lines

- Tobias

#13 lore raw

cover Fix operation on Odroid devices

In reply to Tobias Jakobi

Hello Tobias,

Quoted 18 lines · Tobias Jakobi

Hello,

currently operation on Exynos4412-based Odroid devices is broken.

The bootloader stops with this message:
Card did not respond to voltage select!
*** Warning - MMC init failed, using default environment

This series fixes error handling in the s5p sdhci driver and the cause of the issue, which is an inverted card detection check.

Thanks goes to Marek Vasut and Sjoerd Simons who helped me on IRC to get this triaged.

With best wishes,
Tobias

Quoted 9 lines

Tobias Jakobi (4):
exynos: Properly initialize host_caps in s5p_sdhci_core_init()
exynos: Fix passing of errors in exynos_mmc_init()
exynos: be more verbose in process_nodes()
exynos: fix and cleanup do_sdhci_init()

  drivers/mmc/s5p_sdhci.c | 39 +++++++++++++++++++++++----------------
  1 file changed, 23 insertions(+), 16 deletions(-)

NAK for the series, since the real problem is in device tree parsing, not in s5p_sdhci.c. I will send patches in a moment.

Best regards,

Signature · 4 lines
Przemyslaw Marczak
Samsung R&D Institute Poland
Samsung Electronics
p.marczak at samsung.com