Skip to contentSkip to conversation

staging: fbtft: reduce stack usage by avoiding large write_reg() varargs

PATCH In review Sun Jian 3 participants 13 messages Jan 4 – Jan 6

Subsystems (3)
lore
Share & tools
lore

Sun Jian

author
#0 lore raw

cover staging: fbtft: reduce stack usage by avoiding large write_reg() varargs

Hi,

This series fixes clang -Wframe-larger-than=1024 warnings in the fbtft staging drivers.

The warnings are triggered by very large write_reg()/write_register() varargs calls, which result in excessive stack usage.

Switch the affected paths to send a u8 command byte followed by the u8 payload using fbtft_write_buf_dc(). The register values and ordering are kept unchanged; only the transfer method is updated.

Patches:
1/4 staging: fbtft: core: avoid large stack usage in DT init parsing 2/4 staging: fbtft: ssd1351: send gamma table via fbtft_write_buf_dc() 3/4 staging: fbtft: ssd1331: send gamma table via fbtft_write_buf_dc() 4/4 staging: fbtft: hx8353d: send LUT via buffer to reduce stack usage

Thanks,
Sun Jian

--
Sun Jian (4):
staging: fbtft: core: avoid large stack usage in DT init parsing
staging: fbtft: ssd1351: send gamma table via fbtft_write_buf_dc()
staging: fbtft: ssd1331: send gamma table via fbtft_write_buf_dc()
staging: fbtft: hx8353d: send LUT via buffer to reduce stack usage

 drivers/staging/fbtft/fb_hx8353d.c | 38 ++++++++++++++++++++++--------
 drivers/staging/fbtft/fb_ssd1331.c | 29 ++++++++++++-----------
 drivers/staging/fbtft/fb_ssd1351.c | 35 ++++++++++++---------------
 drivers/staging/fbtft/fbtft-core.c | 32 ++++++++++---------------
 4 files changed, 71 insertions(+), 63 deletions(-)
Signature · 1 line
2.43.0
author
#1 lore raw

PATCH 1/4 staging: fbtft: core: avoid large stack usage in DT init parsing

Clang reports a large stack frame for fbtft_init_display_from_property() (-Wframe-larger-than=1024) when the init sequence is emitted through a fixed 64-argument write_register() call.

write_reg()/write_register() relies on NUMARGS((int[]){...}) and large varargs which inflates stack usage. Switch the DT "init" path to send the command byte and the payload via fbtft_write_buf_dc() instead.

No functional change intended: the same register values are sent in the same order, only the transport is changed.

---
 drivers/staging/fbtft/fbtft-core.c | 32 ++++++++++++------------------
 1 file changed, 13 insertions(+), 19 deletions(-)
drivers/staging/fbtft/fbtft-core.c +13 -19
index 8a5ccc8ae0a1..127d0de87e03 100644
@@ -846,7 +846,8 @@ EXPORT_SYMBOL(fbtft_unregister_framebuffer);
 static int fbtft_init_display_from_property(struct fbtft_par *par)
 {
 	struct device *dev = par->info->device;
-	int buf[64], count, index, i, j, ret;
+	u8 buf[64];
+	int count, index, i, j, ret;
 	u32 *values;
 	u32 val;
 
@@ -881,7 +882,7 @@ static int fbtft_init_display_from_property(struct fbtft_par *par)
 					ret = -EINVAL;
 					goto out_free;
 				}
-				buf[i++] = val;
+				buf[i++] = val & 0xFF;
 				val = values[++index];
 			}
 			/* make debug message */
@@ -891,23 +892,16 @@ static int fbtft_init_display_from_property(struct fbtft_par *par)
 				fbtft_par_dbg(DEBUG_INIT_DISPLAY, par,
 					      "buf[%d] = %02X\n", j, buf[j]);
 
-			par->fbtftops.write_register(par, i,
-				buf[0], buf[1], buf[2], buf[3],
-				buf[4], buf[5], buf[6], buf[7],
-				buf[8], buf[9], buf[10], buf[11],
-				buf[12], buf[13], buf[14], buf[15],
-				buf[16], buf[17], buf[18], buf[19],
-				buf[20], buf[21], buf[22], buf[23],
-				buf[24], buf[25], buf[26], buf[27],
-				buf[28], buf[29], buf[30], buf[31],
-				buf[32], buf[33], buf[34], buf[35],
-				buf[36], buf[37], buf[38], buf[39],
-				buf[40], buf[41], buf[42], buf[43],
-				buf[44], buf[45], buf[46], buf[47],
-				buf[48], buf[49], buf[50], buf[51],
-				buf[52], buf[53], buf[54], buf[55],
-				buf[56], buf[57], buf[58], buf[59],
-				buf[60], buf[61], buf[62], buf[63]);
+			/* buf[0] is command, buf[1..i-1] is data */
+			ret = fbtft_write_buf_dc(par, &buf[0], 1, 0);
+			if (ret < 0)
+				goto out_free;
+
+			if (i > 1) {
+				ret = fbtft_write_buf_dc(par, &buf[1], i - 1, 1);
+				if (ret < 0)
+					goto out_free;
+			}
 		} else if (val & FBTFT_OF_INIT_DELAY) {
 			fbtft_par_dbg(DEBUG_INIT_DISPLAY, par,
 				      "init: msleep(%u)\n", val & 0xFFFF);
Signature · 1 line
2.43.0
#2 lore raw

Re: [PATCH v1 1/4] staging: fbtft: core: avoid large stack usage in DT init parsing

Quoted 10 lines · Sun Jian

Clang reports a large stack frame for fbtft_init_display_from_property() (-Wframe-larger-than=1024) when the init sequence is emitted through a fixed 64-argument write_register() call.

write_reg()/write_register() relies on NUMARGS((int[]){...}) and large varargs which inflates stack usage. Switch the DT "init" path to send the command byte and the payload via fbtft_write_buf_dc() instead.

No functional change intended: the same register values are sent in the same order, only the transport is changed.

How did you test this?

...

drivers/staging/fbtft/fbtft-core.c EXPORT_SYMBOL(fbtft_unregister_framebuffer); · line 850

Quoted 4 lines
drivers/staging/fbtft/fbtft-core.c +2 -1
 	struct device *dev = par->info->device;
-	int buf[64], count, index, i, j, ret;
+	u8 buf[64];
+	int count, index, i, j, ret;

Please, try to preserve reversed xmas tree order.

Quoted 3 lines
drivers/staging/fbtft/fbtft-core.c +0 -0
 	u32 *values;
 	u32 val;

...

drivers/staging/fbtft/fbtft-core.c static int fbtft_init_display_from_property(struct fbtft_par *par) · line 885

Quoted 2 lines
drivers/staging/fbtft/fbtft-core.c +1 -1
-				buf[i++] = val;
+				buf[i++] = val & 0xFF;

Unneeded change, I suppose.

...

Quoted 17 lines
drivers/staging/fbtft/fbtft-core.c +0 -17
-			par->fbtftops.write_register(par, i,
-				buf[0], buf[1], buf[2], buf[3],
-				buf[4], buf[5], buf[6], buf[7],
-				buf[8], buf[9], buf[10], buf[11],
-				buf[12], buf[13], buf[14], buf[15],
-				buf[16], buf[17], buf[18], buf[19],
-				buf[20], buf[21], buf[22], buf[23],
-				buf[24], buf[25], buf[26], buf[27],
-				buf[28], buf[29], buf[30], buf[31],
-				buf[32], buf[33], buf[34], buf[35],
-				buf[36], buf[37], buf[38], buf[39],
-				buf[40], buf[41], buf[42], buf[43],
-				buf[44], buf[45], buf[46], buf[47],
-				buf[48], buf[49], buf[50], buf[51],
-				buf[52], buf[53], buf[54], buf[55],
-				buf[56], buf[57], buf[58], buf[59],
-				buf[60], buf[61], buf[62], buf[63]);

drivers/staging/fbtft/fbtft-core.c static int fbtft_init_display_from_property(struct fbtft_par *par) · line 904

Quoted 10 lines
drivers/staging/fbtft/fbtft-core.c +10 -0
+			/* buf[0] is command, buf[1..i-1] is data */
+			ret = fbtft_write_buf_dc(par, &buf[0], 1, 0);
+			if (ret < 0)
+				goto out_free;
+
+			if (i > 1) {
+				ret = fbtft_write_buf_dc(par, &buf[1], i - 1, 1);
+				if (ret < 0)
+					goto out_free;
+			}

I believe this is incorrect change and has not to be applied. write != write_register. Without any evidence of testing, definite NAK to it. Otherwise, please provide detailed testing pattern and which devices were tested.

Signature · 2 lines
With Best Regards,
Andy Shevchenko
author
#3 lore raw

Re: [PATCH v1 1/4] staging: fbtft: core: avoid large stack usage in DT init parsing

Hi Andy,

Thanks for the feedback.

You are right: changing the DT init path from write_register() to fbtft_write_buf_dc() implicitly assumes "cmd byte + payload bytes" and does not preserve the generic write_register() semantics (e.g. regwidth / bus-specific handling).I only have clang/arm64 build coverage (no access to the actual panels),
so I can’t provide runtime validation yet. For the remaining 3 driver-local patches, all affected drivers have .regwidth = 8 and the sequences are “1-byte command + N bytes data” (gamma/LUT). The intent was to avoid the huge write_reg() varargs call that triggers -Wframe-larger-than=1024.

Given the lack of hardware, would you prefer one of the following?

1. Drop the driver changes and instead bump -Wframe-larger-than for these specific objects in the Makefile as an exception; or

2. Keep the driver changes but I should provide a detailed test pattern / list of tested devices — if so, what level of detail would be acceptable (exact panel model + wiring/bus type + expected output), and is “build-only” ever sufficient for warning-only changes in fbtft?

Happy to follow the approach you think is appropriate for this staging driver.

Best regards,
Sun Jian

Quoted 75 lines · Andy Shevchenko
Quoted 10 lines · Sun Jian

Clang reports a large stack frame for fbtft_init_display_from_property() (-Wframe-larger-than=1024) when the init sequence is emitted through a fixed 64-argument write_register() call.

write_reg()/write_register() relies on NUMARGS((int[]){...}) and large varargs which inflates stack usage. Switch the DT "init" path to send the command byte and the payload via fbtft_write_buf_dc() instead.

No functional change intended: the same register values are sent in the same order, only the transport is changed.

How did you test this?

...

drivers/staging/fbtft/fbtft-core.c EXPORT_SYMBOL(fbtft_unregister_framebuffer); · line 850

Quoted 4 lines
      struct device *dev = par->info->device;
drivers/staging/fbtft/fbtft-core.c +2 -1
-     int buf[64], count, index, i, j, ret;
+     u8 buf[64];
+     int count, index, i, j, ret;

Please, try to preserve reversed xmas tree order.

Quoted 3 lines
      u32 *values;
      u32 val;

...

drivers/staging/fbtft/fbtft-core.c static int fbtft_init_display_from_property(struct fbtft_par *par) · line 885

Quoted 2 lines
drivers/staging/fbtft/fbtft-core.c +1 -1
-                             buf[i++] = val;
+                             buf[i++] = val & 0xFF;

Unneeded change, I suppose.

...

Quoted 17 lines
drivers/staging/fbtft/fbtft-core.c +0 -17
-                     par->fbtftops.write_register(par, i,
-                             buf[0], buf[1], buf[2], buf[3],
-                             buf[4], buf[5], buf[6], buf[7],
-                             buf[8], buf[9], buf[10], buf[11],
-                             buf[12], buf[13], buf[14], buf[15],
-                             buf[16], buf[17], buf[18], buf[19],
-                             buf[20], buf[21], buf[22], buf[23],
-                             buf[24], buf[25], buf[26], buf[27],
-                             buf[28], buf[29], buf[30], buf[31],
-                             buf[32], buf[33], buf[34], buf[35],
-                             buf[36], buf[37], buf[38], buf[39],
-                             buf[40], buf[41], buf[42], buf[43],
-                             buf[44], buf[45], buf[46], buf[47],
-                             buf[48], buf[49], buf[50], buf[51],
-                             buf[52], buf[53], buf[54], buf[55],
-                             buf[56], buf[57], buf[58], buf[59],
-                             buf[60], buf[61], buf[62], buf[63]);

drivers/staging/fbtft/fbtft-core.c static int fbtft_init_display_from_property(struct fbtft_par *par) · line 904

Quoted 10 lines
drivers/staging/fbtft/fbtft-core.c +10 -0
+                     /* buf[0] is command, buf[1..i-1] is data */
+                     ret = fbtft_write_buf_dc(par, &buf[0], 1, 0);
+                     if (ret < 0)
+                             goto out_free;
+
+                     if (i > 1) {
+                             ret = fbtft_write_buf_dc(par, &buf[1], i - 1, 1);
+                             if (ret < 0)
+                                     goto out_free;
+                     }

I believe this is incorrect change and has not to be applied. write != write_register. Without any evidence of testing, definite NAK to it. Otherwise, please provide detailed testing pattern and which devices were tested.

--
With Best Regards,
Andy Shevchenko

#4 lore raw

Re: [PATCH v1 1/4] staging: fbtft: core: avoid large stack usage in DT init parsing

Quoted 1 line · sun jian

Thanks for the feedback.

You're welcome, but please, do not top-post!

Quoted 11 lines

You are right: changing the DT init path from write_register() to fbtft_write_buf_dc() implicitly assumes "cmd byte + payload bytes" and does not preserve the generic write_register() semantics (e.g. regwidth / bus-specific handling).I only have clang/arm64 build coverage (no access to the actual panels),
so I can’t provide runtime validation yet. For the remaining 3 driver-local patches, all affected drivers have .regwidth = 8 and the sequences are “1-byte command + N bytes data” (gamma/LUT). The intent was to avoid the huge write_reg() varargs call that triggers -Wframe-larger-than=1024.

Given the lack of hardware, would you prefer one of the following?

How can you test without hardware at hand?

Quoted 9 lines

1. Drop the driver changes and instead bump -Wframe-larger-than for these specific objects in the Makefile as an exception; or

2. Keep the driver changes but I should provide a detailed test pattern / list of tested devices — if so, what level of detail would be acceptable (exact panel model + wiring/bus type + expected output), and is “build-only” ever sufficient for warning-only changes in fbtft?

Happy to follow the approach you think is appropriate for this staging driver.

I already explained in the response to the cover letter. Please, read it.

Quoted 16 lines
Quoted 14 lines · Andy Shevchenko
Quoted 10 lines · Sun Jian

Clang reports a large stack frame for fbtft_init_display_from_property() (-Wframe-larger-than=1024) when the init sequence is emitted through a fixed 64-argument write_register() call.

write_reg()/write_register() relies on NUMARGS((int[]){...}) and large varargs which inflates stack usage. Switch the DT "init" path to send the command byte and the payload via fbtft_write_buf_dc() instead.

No functional change intended: the same register values are sent in the same order, only the transport is changed.

How did you test this?

...

Quoted 32 lines
Quoted 32 lines
Quoted 17 lines
drivers/staging/fbtft/fbtft-core.c +0 -17
-                     par->fbtftops.write_register(par, i,
-                             buf[0], buf[1], buf[2], buf[3],
-                             buf[4], buf[5], buf[6], buf[7],
-                             buf[8], buf[9], buf[10], buf[11],
-                             buf[12], buf[13], buf[14], buf[15],
-                             buf[16], buf[17], buf[18], buf[19],
-                             buf[20], buf[21], buf[22], buf[23],
-                             buf[24], buf[25], buf[26], buf[27],
-                             buf[28], buf[29], buf[30], buf[31],
-                             buf[32], buf[33], buf[34], buf[35],
-                             buf[36], buf[37], buf[38], buf[39],
-                             buf[40], buf[41], buf[42], buf[43],
-                             buf[44], buf[45], buf[46], buf[47],
-                             buf[48], buf[49], buf[50], buf[51],
-                             buf[52], buf[53], buf[54], buf[55],
-                             buf[56], buf[57], buf[58], buf[59],
-                             buf[60], buf[61], buf[62], buf[63]);

drivers/staging/fbtft/fbtft-core.c static int fbtft_init_display_from_property(struct fbtft_par *par) · line 904

Quoted 10 lines
drivers/staging/fbtft/fbtft-core.c +10 -0
+                     /* buf[0] is command, buf[1..i-1] is data */
+                     ret = fbtft_write_buf_dc(par, &buf[0], 1, 0);
+                     if (ret < 0)
+                             goto out_free;
+
+                     if (i > 1) {
+                             ret = fbtft_write_buf_dc(par, &buf[1], i - 1, 1);
+                             if (ret < 0)
+                                     goto out_free;
+                     }

I believe this is incorrect change and has not to be applied. write != write_register. Without any evidence of testing, definite NAK to it. Otherwise, please provide detailed testing pattern and which devices were tested.

Signature · 2 lines
With Best Regards,
Andy Shevchenko
author
#5 lore raw

Re: [PATCH v1 1/4] staging: fbtft: core: avoid large stack usage in DT init parsing

Quoted 6 lines · Andy Shevchenko
Quoted 1 line · sun jian

Thanks for the feedback.

You're welcome, but please, do not top-post!

Sorry about that - I'll use inline replies.

Quoted 7 lines

How can you test without hardware at hand?

I already explained in the response to the cover letter. Please, read it.

Given that, I will drop all the changes.

Thanks,
Sun Jian

author
#6 lore raw

PATCH 2/4 staging: fbtft: ssd1351: send gamma table via fbtft_write_buf_dc()

In reply to Sun Jian

Clang reports a large stack frame in set_gamma() (-Wframe-larger-than=1024) due to the large write_reg() call emitting 63 gamma bytes via varargs.

Send the command byte (0xB8) and the gamma payload using fbtft_write_buf_dc() to avoid the varargs/NUMARGS stack blow-up.

No functional change intended.

---
 drivers/staging/fbtft/fb_ssd1351.c | 35 +++++++++++++-----------------
 1 file changed, 15 insertions(+), 20 deletions(-)
drivers/staging/fbtft/fb_ssd1351.c +15 -20
index 6736b09b2f45..b4ab2c81e528 100644
@@ -119,43 +119,38 @@ static int set_var(struct fbtft_par *par)
  */
 static int set_gamma(struct fbtft_par *par, u32 *curves)
 {
-	unsigned long tmp[GAMMA_NUM * GAMMA_LEN];
+	u8 data[GAMMA_LEN];
+	u8 cmd = 0xB8;
 	int i, acc = 0;
+	int ret;
 
-	for (i = 0; i < 63; i++) {
+	for (i = 0; i < GAMMA_LEN; i++) {
 		if (i > 0 && curves[i] < 2) {
 			dev_err(par->info->device,
 				"Illegal value in Grayscale Lookup Table at index %d : %d. Must be greater than 1\n",
 				i, curves[i]);
 			return -EINVAL;
 		}
+
 		acc += curves[i];
-		tmp[i] = acc;
+
 		if (acc > 180) {
 			dev_err(par->info->device,
 				"Illegal value(s) in Grayscale Lookup Table. At index=%d : %d, the accumulated value has exceeded 180\n",
 				i, acc);
 			return -EINVAL;
 		}
+
+		data[i] = acc;
 	}
 
-	write_reg(par, 0xB8,
-		  tmp[0],  tmp[1],  tmp[2],  tmp[3],
-		  tmp[4],  tmp[5],  tmp[6],  tmp[7],
-		  tmp[8],  tmp[9],  tmp[10], tmp[11],
-		  tmp[12], tmp[13], tmp[14], tmp[15],
-		  tmp[16], tmp[17], tmp[18], tmp[19],
-		  tmp[20], tmp[21], tmp[22], tmp[23],
-		  tmp[24], tmp[25], tmp[26], tmp[27],
-		  tmp[28], tmp[29], tmp[30], tmp[31],
-		  tmp[32], tmp[33], tmp[34], tmp[35],
-		  tmp[36], tmp[37], tmp[38], tmp[39],
-		  tmp[40], tmp[41], tmp[42], tmp[43],
-		  tmp[44], tmp[45], tmp[46], tmp[47],
-		  tmp[48], tmp[49], tmp[50], tmp[51],
-		  tmp[52], tmp[53], tmp[54], tmp[55],
-		  tmp[56], tmp[57], tmp[58], tmp[59],
-		  tmp[60], tmp[61], tmp[62]);
+	ret = fbtft_write_buf_dc(par, &cmd, 1, 0);
+	if (ret < 0)
+		return ret;
+
+	ret = fbtft_write_buf_dc(par, data, sizeof(data), 1);
+	if (ret < 0)
+		return ret;
 
 	return 0;
 }
Signature · 1 line
2.43.0
#7 lore raw

Re: [PATCH v1 2/4] staging: fbtft: ssd1351: send gamma table via fbtft_write_buf_dc()

Quoted 18 lines · Sun Jian

Clang reports a large stack frame in set_gamma() (-Wframe-larger-than=1024) due to the large write_reg() call emitting 63 gamma bytes via varargs.

Send the command byte (0xB8) and the gamma payload using fbtft_write_buf_dc() to avoid the varargs/NUMARGS stack blow-up.

No functional change intended.

---
 drivers/staging/fbtft/fb_ssd1351.c | 35 +++++++++++++-----------------
 1 file changed, 15 insertions(+), 20 deletions(-)
drivers/staging/fbtft/fb_ssd1351.c +0 -0
index 6736b09b2f45..b4ab2c81e528 100644

drivers/staging/fbtft/fb_ssd1351.c static int set_var(struct fbtft_par *par) · line 122

Quoted 5 lines
drivers/staging/fbtft/fb_ssd1351.c +1 -1
@@ -119,43 +119,38 @@ static int set_var(struct fbtft_par *par)
  */
 static int set_gamma(struct fbtft_par *par, u32 *curves)
 {
-	unsigned long tmp[GAMMA_NUM * GAMMA_LEN];
+	u8 data[GAMMA_LEN];

Ugh... GAMMA_NUM is 1 so this is an annoying calculation. So what this does is it changes the type from unsigned long to u8 and renames the variable. I am fine with renaming the variable it's unrelated and makes the review harder.

drivers/staging/fbtft/fb_ssd1351.c static int set_var(struct fbtft_par *par) · line 127

Quoted 6 lines
drivers/staging/fbtft/fb_ssd1351.c +3 -1
+	u8 cmd = 0xB8;
 	int i, acc = 0;
+	int ret;
 
-	for (i = 0; i < 63; i++) {
+	for (i = 0; i < GAMMA_LEN; i++) {

GAMMA_LEN is 63. So this looks like a change, but it's an unrelated cleanup.

drivers/staging/fbtft/fb_ssd1351.c static int set_var(struct fbtft_par *par) · line 134

Quoted 7 lines
drivers/staging/fbtft/fb_ssd1351.c +1 -0
 		if (i > 0 && curves[i] < 2) {
 			dev_err(par->info->device,
 				"Illegal value in Grayscale Lookup Table at index %d : %d. Must be greater than 1\n",
 				i, curves[i]);
 			return -EINVAL;
 		}
+

This is an unrelated white space change.

drivers/staging/fbtft/fb_ssd1351.c static int set_var(struct fbtft_par *par) · line 144

Quoted 11 lines
drivers/staging/fbtft/fb_ssd1351.c +3 -1
 		acc += curves[i];
-		tmp[i] = acc;
+
 		if (acc > 180) {
 			dev_err(par->info->device,
 				"Illegal value(s) in Grayscale Lookup Table. At index=%d : %d, the accumulated value has exceeded 180\n",
 				i, acc);
 			return -EINVAL;
 		}
+
+		data[i] = acc;

Here we move the acc assignment after the sanity check, but it's just an unrelated cleanup.

Quoted 16 lines
drivers/staging/fbtft/fb_ssd1351.c +0 -14
 	}
 
-	write_reg(par, 0xB8,
-		  tmp[0],  tmp[1],  tmp[2],  tmp[3],
-		  tmp[4],  tmp[5],  tmp[6],  tmp[7],
-		  tmp[8],  tmp[9],  tmp[10], tmp[11],
-		  tmp[12], tmp[13], tmp[14], tmp[15],
-		  tmp[16], tmp[17], tmp[18], tmp[19],
-		  tmp[20], tmp[21], tmp[22], tmp[23],
-		  tmp[24], tmp[25], tmp[26], tmp[27],
-		  tmp[28], tmp[29], tmp[30], tmp[31],
-		  tmp[32], tmp[33], tmp[34], tmp[35],
-		  tmp[36], tmp[37], tmp[38], tmp[39],
-		  tmp[40], tmp[41], tmp[42], tmp[43],
-		  tmp[44], tmp[45], tmp[46], tmp[47],
-		  tmp[48], tmp[49], tmp[50], tmp[51],

drivers/staging/fbtft/fb_ssd1351.c static int set_var(struct fbtft_par *par) · line 153

Quoted 10 lines
drivers/staging/fbtft/fb_ssd1351.c +7 -3
-		  tmp[52], tmp[53], tmp[54], tmp[55],
-		  tmp[56], tmp[57], tmp[58], tmp[59],
-		  tmp[60], tmp[61], tmp[62]);
+	ret = fbtft_write_buf_dc(par, &cmd, 1, 0);
+	if (ret < 0)
+		return ret;
+
+	ret = fbtft_write_buf_dc(par, data, sizeof(data), 1);
+	if (ret < 0)
+		return ret;

These are good changes. Just change the type from unsigned long to u8 and use fbtft_write_buf_dc() instead of write_reg(). Then do the other changes in a separate patch.

Same for the other patches.

regards,
dan carpenter

Quoted 6 lines
drivers/staging/fbtft/fb_ssd1351.c +0 -0
 	return 0;
 }
Signature · 1 line
2.43.0
author
#8 lore raw

Re: [PATCH v1 2/4] staging: fbtft: ssd1351: send gamma table via fbtft_write_buf_dc()

Hi Dan,

Thanks. Agreed.

For v2 I’ll keep this patch minimal: only switch from write_reg() varargs to fbtft_write_buf_dc() and reduce the stack usage, without other changes.

Any follow-up cleanups (if needed) will be sent as a separate patch.

Thanks,
Sun Jian

Quoted 108 lines · Dan Carpenter
Quoted 18 lines · Sun Jian

Clang reports a large stack frame in set_gamma() (-Wframe-larger-than=1024) due to the large write_reg() call emitting 63 gamma bytes via varargs.

Send the command byte (0xB8) and the gamma payload using fbtft_write_buf_dc() to avoid the varargs/NUMARGS stack blow-up.

No functional change intended.

---
 drivers/staging/fbtft/fb_ssd1351.c | 35 +++++++++++++-----------------
 1 file changed, 15 insertions(+), 20 deletions(-)
drivers/staging/fbtft/fb_ssd1351.c +0 -0
index 6736b09b2f45..b4ab2c81e528 100644

drivers/staging/fbtft/fb_ssd1351.c static int set_var(struct fbtft_par *par) · line 122

Quoted 5 lines
drivers/staging/fbtft/fb_ssd1351.c +1 -1
@@ -119,43 +119,38 @@ static int set_var(struct fbtft_par *par)
  */
 static int set_gamma(struct fbtft_par *par, u32 *curves)
 {
-     unsigned long tmp[GAMMA_NUM * GAMMA_LEN];
+     u8 data[GAMMA_LEN];

Ugh... GAMMA_NUM is 1 so this is an annoying calculation. So what this does is it changes the type from unsigned long to u8 and renames the variable. I am fine with renaming the variable it's unrelated and makes the review harder.

drivers/staging/fbtft/fb_ssd1351.c static int set_var(struct fbtft_par *par) · line 127

Quoted 6 lines
drivers/staging/fbtft/fb_ssd1351.c +3 -1
+     u8 cmd = 0xB8;
      int i, acc = 0;
+     int ret;
 
-     for (i = 0; i < 63; i++) {
+     for (i = 0; i < GAMMA_LEN; i++) {

GAMMA_LEN is 63. So this looks like a change, but it's an unrelated cleanup.

drivers/staging/fbtft/fb_ssd1351.c static int set_var(struct fbtft_par *par) · line 134

Quoted 7 lines
drivers/staging/fbtft/fb_ssd1351.c +1 -0
              if (i > 0 && curves[i] < 2) {
                      dev_err(par->info->device,
                              "Illegal value in Grayscale Lookup Table at index %d : %d. Must be greater than 1\n",
                              i, curves[i]);
                      return -EINVAL;
              }
+

This is an unrelated white space change.

drivers/staging/fbtft/fb_ssd1351.c static int set_var(struct fbtft_par *par) · line 144

Quoted 11 lines
drivers/staging/fbtft/fb_ssd1351.c +3 -1
              acc += curves[i];
-             tmp[i] = acc;
+
              if (acc > 180) {
                      dev_err(par->info->device,
                              "Illegal value(s) in Grayscale Lookup Table. At index=%d : %d, the accumulated value has exceeded 180\n",
                              i, acc);
                      return -EINVAL;
              }
+
+             data[i] = acc;

Here we move the acc assignment after the sanity check, but it's just an unrelated cleanup.

Quoted 16 lines
drivers/staging/fbtft/fb_ssd1351.c +0 -14
      }
 
-     write_reg(par, 0xB8,
-               tmp[0],  tmp[1],  tmp[2],  tmp[3],
-               tmp[4],  tmp[5],  tmp[6],  tmp[7],
-               tmp[8],  tmp[9],  tmp[10], tmp[11],
-               tmp[12], tmp[13], tmp[14], tmp[15],
-               tmp[16], tmp[17], tmp[18], tmp[19],
-               tmp[20], tmp[21], tmp[22], tmp[23],
-               tmp[24], tmp[25], tmp[26], tmp[27],
-               tmp[28], tmp[29], tmp[30], tmp[31],
-               tmp[32], tmp[33], tmp[34], tmp[35],
-               tmp[36], tmp[37], tmp[38], tmp[39],
-               tmp[40], tmp[41], tmp[42], tmp[43],
-               tmp[44], tmp[45], tmp[46], tmp[47],
-               tmp[48], tmp[49], tmp[50], tmp[51],

drivers/staging/fbtft/fb_ssd1351.c static int set_var(struct fbtft_par *par) · line 153

Quoted 10 lines
drivers/staging/fbtft/fb_ssd1351.c +7 -3
-               tmp[52], tmp[53], tmp[54], tmp[55],
-               tmp[56], tmp[57], tmp[58], tmp[59],
-               tmp[60], tmp[61], tmp[62]);
+     ret = fbtft_write_buf_dc(par, &cmd, 1, 0);
+     if (ret < 0)
+             return ret;
+
+     ret = fbtft_write_buf_dc(par, data, sizeof(data), 1);
+     if (ret < 0)
+             return ret;

These are good changes. Just change the type from unsigned long to u8 and use fbtft_write_buf_dc() instead of write_reg(). Then do the other changes in a separate patch.

Same for the other patches.

regards,
dan carpenter

Quoted 6 lines
drivers/staging/fbtft/fb_ssd1351.c +0 -1
      return 0;
 }
--

2.43.0

author
#9 lore raw

PATCH 3/4 staging: fbtft: ssd1331: send gamma table via fbtft_write_buf_dc()

In reply to Sun Jian

Clang reports a large stack frame in set_gamma() (-Wframe-larger-than=1024) because write_reg() expands into a large varargs call and triggers NUMARGS((int[]){...}) stack allocation.

Use fbtft_write_buf_dc() to send the command byte (0xB8) and the gamma payload as a buffer.

No functional change intended.

---
 drivers/staging/fbtft/fb_ssd1331.c | 29 +++++++++++++++--------------
 1 file changed, 15 insertions(+), 14 deletions(-)
drivers/staging/fbtft/fb_ssd1331.c +15 -14
index 06b7056d6c71..cbe10f191f5b 100644
@@ -130,37 +130,38 @@ static void write_reg8_bus8(struct fbtft_par *par, int len, ...)
  */
 static int set_gamma(struct fbtft_par *par, u32 *curves)
 {
-	unsigned long tmp[GAMMA_NUM * GAMMA_LEN];
+	u8 data[GAMMA_LEN];
+	u8 cmd = 0xB8;
 	int i, acc = 0;
+	int ret;
 
-	for (i = 0; i < 63; i++) {
+	for (i = 0; i < GAMMA_LEN; i++) {
 		if (i > 0 && curves[i] < 2) {
 			dev_err(par->info->device,
 				"Illegal value in Grayscale Lookup Table at index %d. Must be greater than 1\n",
 				i);
 			return -EINVAL;
 		}
+
 		acc += curves[i];
-		tmp[i] = acc;
+
 		if (acc > 180) {
 			dev_err(par->info->device,
 				"Illegal value(s) in Grayscale Lookup Table. At index=%d, the accumulated value has exceeded 180\n",
 				i);
 			return -EINVAL;
 		}
+
+		data[i] = acc;
 	}
 
-	write_reg(par, 0xB8,
-		  tmp[0], tmp[1], tmp[2], tmp[3], tmp[4], tmp[5], tmp[6],
-		  tmp[7], tmp[8], tmp[9], tmp[10], tmp[11], tmp[12], tmp[13],
-		  tmp[14], tmp[15], tmp[16], tmp[17], tmp[18], tmp[19], tmp[20],
-		  tmp[21], tmp[22], tmp[23], tmp[24], tmp[25], tmp[26],	tmp[27],
-		  tmp[28], tmp[29], tmp[30], tmp[31], tmp[32], tmp[33], tmp[34],
-		  tmp[35], tmp[36], tmp[37], tmp[38], tmp[39], tmp[40], tmp[41],
-		  tmp[42], tmp[43], tmp[44], tmp[45], tmp[46], tmp[47], tmp[48],
-		  tmp[49], tmp[50], tmp[51], tmp[52], tmp[53], tmp[54], tmp[55],
-		  tmp[56], tmp[57], tmp[58], tmp[59], tmp[60], tmp[61],
-		  tmp[62]);
+	ret = fbtft_write_buf_dc(par, &cmd, 1, 0);
+	if (ret < 0)
+		return ret;
+
+	ret = fbtft_write_buf_dc(par, data, sizeof(data), 1);
+	if (ret < 0)
+		return ret;
 
 	return 0;
 }
Signature · 1 line
2.43.0
author
#10 lore raw

PATCH 4/4 staging: fbtft: hx8353d: send LUT via buffer to reduce stack usage

In reply to Sun Jian

Clang reports a large stack frame in init_display()
(-Wframe-larger-than=1024) due to the very large write_reg(MIPI_DCS_WRITE_LUT, ...) call.

Send MIPI_DCS_WRITE_LUT followed by the LUT payload using fbtft_write_buf_dc() to avoid the varargs/NUMARGS stack blow-up.

No functional change intended.

---
 drivers/staging/fbtft/fb_hx8353d.c | 38 ++++++++++++++++++++++--------
 1 file changed, 28 insertions(+), 10 deletions(-)
drivers/staging/fbtft/fb_hx8353d.c +28 -10
index 3e73b69b6a27..7e667d60792f 100644
@@ -17,6 +17,21 @@
 #define DRVNAME "fb_hx8353d"
 #define DEFAULT_GAMMA "50 77 40 08 BF 00 03 0F 00 01 73 00 72 03 B0 0F 08 00 0F"
 
+static const u8 lut[] = {
+			  0,  2,  4,  6,  8, 10, 12, 14, 16, 18, 20, 22, 24, 26, 28, 30,
+			 32, 34, 36, 38, 40, 42, 44, 46, 48, 50, 52, 54, 56, 58, 60, 62,
+			  0,  1,  2,  3,  4,  5,  6,  7,  8,  9, 10, 11, 12, 13, 14, 15,
+			 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31,
+			 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47,
+			 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63,
+			  0,  2,  4,  6,  8, 10, 12, 14, 16, 18, 20, 22, 24, 26, 28, 30,
+			 32, 34, 36, 38, 40, 42, 44, 46, 48, 50, 52, 54, 56, 58, 60, 62,
+			  0,  1,  2,  3,  4,  5,  6,  7,  8,  9, 10, 11, 12, 13, 14, 15,
+			 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31,
+			 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47,
+			 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63,
+		};
+
 static int init_display(struct fbtft_par *par)
 {
 	par->fbtftops.reset(par);
@@ -48,18 +63,21 @@ static int init_display(struct fbtft_par *par)
 	write_reg(par, MIPI_DCS_SET_DISPLAY_ON);
 
 	/* RGBSET */
-	write_reg(par, MIPI_DCS_WRITE_LUT,
-		  0,  2,  4,  6,  8, 10, 12, 14, 16, 18, 20, 22, 24, 26, 28, 30,
-		32, 34, 36, 38, 40, 42, 44, 46, 48, 50, 52, 54, 56, 58, 60, 62,
-		 0,  1,  2,  3,  4,  5,  6,  7,  8,  9, 10, 11, 12, 13, 14, 15,
-		16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31,
-		32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47,
-		48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63,
-		 0,  2,  4,  6,  8, 10, 12, 14, 16, 18, 20, 22, 24, 26, 28, 30,
-		32, 34, 36, 38, 40, 42, 44, 46, 48, 50, 52, 54, 56, 58, 60, 62);
+	{
+		u8 cmd = MIPI_DCS_WRITE_LUT;
+		int ret;
+
+		ret = fbtft_write_buf_dc(par, &cmd, 1, 0);
+		if (ret < 0)
+			return ret;
+
+		ret = fbtft_write_buf_dc(par, (void *)lut, sizeof(lut), 1);
+		if (ret < 0)
+			return ret;
+	}
 
 	return 0;
-};
+}
 
 static void set_addr_win(struct fbtft_par *par, int xs, int ys, int xe, int ye)
 {
Signature · 1 line
2.43.0
#11 lore raw

Re: [PATCH v1 4/4] staging: fbtft: hx8353d: send LUT via buffer to reduce stack usage

Quoted 8 lines · Sun Jian

Clang reports a large stack frame in init_display()
(-Wframe-larger-than=1024) due to the very large write_reg(MIPI_DCS_WRITE_LUT, ...) call.

Send MIPI_DCS_WRITE_LUT followed by the LUT payload using fbtft_write_buf_dc() to avoid the varargs/NUMARGS stack blow-up.

No functional change intended.

...

drivers/staging/fbtft/fb_hx8353d.c line 32

Quoted 13 lines
drivers/staging/fbtft/fb_hx8353d.c +13 -0
+static const u8 lut[] = {
+			  0,  2,  4,  6,  8, 10, 12, 14, 16, 18, 20, 22, 24, 26, 28, 30,
+			 32, 34, 36, 38, 40, 42, 44, 46, 48, 50, 52, 54, 56, 58, 60, 62,
+			  0,  1,  2,  3,  4,  5,  6,  7,  8,  9, 10, 11, 12, 13, 14, 15,
+			 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31,
+			 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47,
+			 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63,
+			  0,  2,  4,  6,  8, 10, 12, 14, 16, 18, 20, 22, 24, 26, 28, 30,
+			 32, 34, 36, 38, 40, 42, 44, 46, 48, 50, 52, 54, 56, 58, 60, 62,
+			  0,  1,  2,  3,  4,  5,  6,  7,  8,  9, 10, 11, 12, 13, 14, 15,
+			 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31,
+			 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47,
+			 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63,

Two tabs too many on each line.

Quoted 2 lines
drivers/staging/fbtft/fb_hx8353d.c +2 -0
+		};
+
Signature · 2 lines
With Best Regards,
Andy Shevchenko
#12 lore raw
In reply to Sun Jian
Quoted 9 lines · Sun Jian

This series fixes clang -Wframe-larger-than=1024 warnings in the fbtft staging drivers.

The warnings are triggered by very large write_reg()/write_register() varargs calls, which result in excessive stack usage.

Switch the affected paths to send a u8 command byte followed by the u8 payload using fbtft_write_buf_dc(). The register values and ordering are kept unchanged; only the transfer method is updated.

Looking at the patches I think this is wrong. W.o. detailed test pattern provided and the list of the devices, NAK.

If you want to address a warning without HW being accessible, perhaps you just need a simple bump in the Makefile as an exception, however it's also doubtful as it will hide a potential issue with the stack in the future.

Signature · 2 lines
With Best Regards,
Andy Shevchenko