Skip to contentSkip to conversation

clk: Don't try to enable critical clocks if prepare failed

PATCH Applied Guenter Roeck 3 participants 6 messages Dec 25, 2019 – Dec 27, 2019 Common clk

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 clk: Don't try to enable critical clocks if prepare failed

The following traceback is seen if a critical clock fails to prepare.

bcm2835-clk 3f101000.cprman: plld: couldn't lock PLL

------------[ cut here ]------------
Enabling unprepared plld_per
WARNING: CPU: 1 PID: 1 at drivers/clk/clk.c:1014 clk_core_enable+0xcc/0x2c0
...
Call trace:
 clk_core_enable+0xcc/0x2c0
 __clk_register+0x5c4/0x788
 devm_clk_hw_register+0x4c/0xb0
 bcm2835_register_pll_divider+0xc0/0x150
 bcm2835_clk_probe+0x134/0x1e8
 platform_drv_probe+0x50/0xa0
 really_probe+0xd4/0x308
 driver_probe_device+0x54/0xe8
 device_driver_attach+0x6c/0x78
 __driver_attach+0x54/0xd8

...

Check return values from clk_core_prepare() and clk_core_enable() and bail out if any of those functions returns an error.

Cc Jerome Brunet <[email protected]>
Fixes 99652a469df1 ("clk: migrate the count of orphaned clocks at init")
---
 drivers/clk/clk.c | 10 ++++++++--
 1 file changed, 8 insertions(+), 2 deletions(-)
drivers/clk/clk.c +8 -2
index 6a11239ccde3..772258de2d1f 100644
@@ -3426,11 +3426,17 @@ static int __clk_core_init(struct clk_core *core)
 	if (core->flags & CLK_IS_CRITICAL) {
 		unsigned long flags;
 
-		clk_core_prepare(core);
+		ret = clk_core_prepare(core);
+		if (ret)
+			goto out;
 
 		flags = clk_enable_lock();
-		clk_core_enable(core);
+		ret = clk_core_enable(core);
 		clk_enable_unlock(flags);
+		if (ret) {
+			clk_core_unprepare(core);
+			goto out;
+		}
 	}
 
 	clk_core_reparent_orphans_nolock();
Signature · 1 line
2.17.1
#1 lore raw
Quoted 41 lines · Guenter Roeck

The following traceback is seen if a critical clock fails to prepare.

bcm2835-clk 3f101000.cprman: plld: couldn't lock PLL

------------[ cut here ]------------
Enabling unprepared plld_per
WARNING: CPU: 1 PID: 1 at drivers/clk/clk.c:1014 clk_core_enable+0xcc/0x2c0
...
Call trace:
 clk_core_enable+0xcc/0x2c0
 __clk_register+0x5c4/0x788
 devm_clk_hw_register+0x4c/0xb0
 bcm2835_register_pll_divider+0xc0/0x150
 bcm2835_clk_probe+0x134/0x1e8
 platform_drv_probe+0x50/0xa0
 really_probe+0xd4/0x308
 driver_probe_device+0x54/0xe8
 device_driver_attach+0x6c/0x78
 __driver_attach+0x54/0xd8

...

Check return values from clk_core_prepare() and clk_core_enable() and bail out if any of those functions returns an error.

Cc Jerome Brunet <[email protected]>
Fixes 99652a469df1 ("clk: migrate the count of orphaned clocks at init")
---
 drivers/clk/clk.c | 10 ++++++++--
 1 file changed, 8 insertions(+), 2 deletions(-)
drivers/clk/clk.c +2 -1
index 6a11239ccde3..772258de2d1f 100644
@@ -3426,11 +3426,17 @@ static int __clk_core_init(struct clk_core *core)
 	if (core->flags & CLK_IS_CRITICAL) {
 		unsigned long flags;
 
-		clk_core_prepare(core);
+		ret = clk_core_prepare(core);
+		if (ret)

drivers/clk/clk.c static int __clk_core_init(struct clk_core *core) · line 3439

Quoted 10 lines
drivers/clk/clk.c +6 -1
+			goto out;
 
 		flags = clk_enable_lock();
-		clk_core_enable(core);
+		ret = clk_core_enable(core);
 		clk_enable_unlock(flags);
+		if (ret) {
+			clk_core_unprepare(core);
+			goto out;
+		}

Hi Guenter,

It looks like it was a mistake to discard the possibility of a failure here. Thanks for correcting this.

However, we would not want a critical clock to silently fail to enable. This might lead to unexpected behavior which are generally hard (and annoying) to debug.

Would you mind adding some kind of warning trace in case this fails ?

Thx

Quoted 3 lines
drivers/clk/clk.c +0 -0
 	}
 
 	clk_core_reparent_orphans_nolock();
#2 lore raw
Quoted 57 lines · Jerome Brunet
Quoted 41 lines · Guenter Roeck

The following traceback is seen if a critical clock fails to prepare.

bcm2835-clk 3f101000.cprman: plld: couldn't lock PLL

------------[ cut here ]------------
Enabling unprepared plld_per
WARNING: CPU: 1 PID: 1 at drivers/clk/clk.c:1014 clk_core_enable+0xcc/0x2c0
...
Call trace:
  clk_core_enable+0xcc/0x2c0
  __clk_register+0x5c4/0x788
  devm_clk_hw_register+0x4c/0xb0
  bcm2835_register_pll_divider+0xc0/0x150
  bcm2835_clk_probe+0x134/0x1e8
  platform_drv_probe+0x50/0xa0
  really_probe+0xd4/0x308
  driver_probe_device+0x54/0xe8
  device_driver_attach+0x6c/0x78
  __driver_attach+0x54/0xd8

...

Check return values from clk_core_prepare() and clk_core_enable() and bail out if any of those functions returns an error.

Cc Jerome Brunet <[email protected]>
Fixes 99652a469df1 ("clk: migrate the count of orphaned clocks at init")
---
  drivers/clk/clk.c | 10 ++++++++--
  1 file changed, 8 insertions(+), 2 deletions(-)
drivers/clk/clk.c +2 -1
index 6a11239ccde3..772258de2d1f 100644
@@ -3426,11 +3426,17 @@ static int __clk_core_init(struct clk_core *core)
  	if (core->flags & CLK_IS_CRITICAL) {
  		unsigned long flags;
 
-		clk_core_prepare(core);
+		ret = clk_core_prepare(core);
+		if (ret)

drivers/clk/clk.c static int __clk_core_init(struct clk_core *core) · line 3439

Quoted 10 lines
drivers/clk/clk.c +6 -1
+			goto out;
 
  		flags = clk_enable_lock();
-		clk_core_enable(core);
+		ret = clk_core_enable(core);
  		clk_enable_unlock(flags);
+		if (ret) {
+			clk_core_unprepare(core);
+			goto out;
+		}

Hi Guenter,

Quoted 9 lines

It looks like it was a mistake to discard the possibility of a failure here. Thanks for correcting this.

However, we would not want a critical clock to silently fail to enable. This might lead to unexpected behavior which are generally hard (and annoying) to debug.

Would you mind adding some kind of warning trace in case this fails ?

The really relevant information is:

bcm2835-clk 3f101000.cprman: plld: couldn't lock PLL

which is already displayed (and not surprising since cprman isn't implemented in qemu). While I agree that an error message might be useful, replacing one traceback with another doesn't really make sense to me, and I am not really a friend of spreading tracebacks throughout the kernel. Please feel free to consider this patch to be a bug report, and feel free to ignore it and suggest something else.

Thanks,
Guenter

maintainer
#3 lore raw

Quoting Guenter Roeck (2019-12-26 09:22:10)

Quoted 11 lines · Jerome Brunet
Quoted 7 lines

However, we would not want a critical clock to silently fail to enable. This might lead to unexpected behavior which are generally hard (and annoying) to debug.

Would you mind adding some kind of warning trace in case this fails ?

The really relevant information is:

Quoted 8 lines

bcm2835-clk 3f101000.cprman: plld: couldn't lock PLL

which is already displayed (and not surprising since cprman isn't implemented in qemu). While I agree that an error message might be useful, replacing one traceback with another doesn't really make sense to me, and I am not really a friend of spreading tracebacks throughout the kernel. Please feel free to consider this patch to be a bug report, and feel free to ignore it and suggest something else.

Can the cprman device node be disabled or removed in the DT that qemu uses? If it isn't actually implemented then it shouldn't be in the DT. Presumably that will make this traceback go away.

#4 lore raw
Quoted 21 lines · Stephen Boyd

Quoting Guenter Roeck (2019-12-26 09:22:10)

Quoted 11 lines · Jerome Brunet
Quoted 7 lines

However, we would not want a critical clock to silently fail to enable. This might lead to unexpected behavior which are generally hard (and annoying) to debug.

Would you mind adding some kind of warning trace in case this fails ?

The really relevant information is:

Quoted 8 lines

bcm2835-clk 3f101000.cprman: plld: couldn't lock PLL

which is already displayed (and not surprising since cprman isn't implemented in qemu). While I agree that an error message might be useful, replacing one traceback with another doesn't really make sense to me, and I am not really a friend of spreading tracebacks throughout the kernel. Please feel free to consider this patch to be a bug report, and feel free to ignore it and suggest something else.

Quoted 4 lines

Can the cprman device node be disabled or removed in the DT that qemu uses? If it isn't actually implemented then it shouldn't be in the DT. Presumably that will make this traceback go away.

cprman feeds all clocks. If the node isn't there, the system doesn't boot. Also, I don't modify devicetree files in my boot tests; that would defeat the purpose - like, in this case, to find missing error handling.

Again, please feel free to ignore this patch.

Guenter

maintainer Applied
#5 lore raw
In reply to Guenter Roeck

Quoting Guenter Roeck (2019-12-25 08:34:29)

Quoted 20 lines

The following traceback is seen if a critical clock fails to prepare.

bcm2835-clk 3f101000.cprman: plld: couldn't lock PLL

------------[ cut here ]------------
Enabling unprepared plld_per
WARNING: CPU: 1 PID: 1 at drivers/clk/clk.c:1014 clk_core_enable+0xcc/0x2c0
...
Call trace:
 clk_core_enable+0xcc/0x2c0
 __clk_register+0x5c4/0x788
 devm_clk_hw_register+0x4c/0xb0
 bcm2835_register_pll_divider+0xc0/0x150
 bcm2835_clk_probe+0x134/0x1e8
 platform_drv_probe+0x50/0xa0
 really_probe+0xd4/0x308
 driver_probe_device+0x54/0xe8
 device_driver_attach+0x6c/0x78
 __driver_attach+0x54/0xd8

...

Quoted 7 lines

Check return values from clk_core_prepare() and clk_core_enable() and bail out if any of those functions returns an error.

Cc Jerome Brunet <[email protected]>
Fixes 99652a469df1 ("clk: migrate the count of orphaned clocks at init")

---

Applied to clk-fixes