gpu: host1x: Do not output error message for deferred probe
PATCH 1/1 gpu: host1x: Do not output error message for deferred probe
From: Thierry Reding <[email protected]>
When deferring probe, avoid logging a confusing error message. While at it, make the error message more informational.
--- drivers/gpu/host1x/dev.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-)
drivers/gpu/host1x/dev.c +4 -1
@@ -247,8 +247,11 @@ static int host1x_probe(struct platform_device *pdev) host->clk =devm_clk_get (&pdev->dev,NULL );if (IS_ERR (host->clk)) { -dev_err (&pdev->dev,"failed to get clock\n" ); err =PTR_ERR (host->clk); + +if (err != -EPROBE_DEFER ) +dev_err (&pdev->dev,"failed to get clock: %d\n" , err); +return err; }
Signature · 6 lines
2.21.0 _______________________________________________ dri-devel mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/dri-devel
04.06.2019 18:31, Thierry Reding пишет:
Quoted 16 lines
From: Thierry Reding <[email protected]>
When deferring probe, avoid logging a confusing error message. While at it, make the error message more informational.
Signed-off-by Thierry Reding <[email protected]>--- drivers/gpu/host1x/dev.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-)drivers/gpu/host1x/dev.c +0
-0@@ -247,8 +247,11 @@ static int host1x_probe(struct platform_device *pdev)
drivers/gpu/host1x/dev.c static int host1x_probe(struct platform_device *pdev) · line 256
Quoted 10 lines
drivers/gpu/host1x/dev.c +4
-1host->clk =devm_clk_get (&pdev->dev,NULL );if (IS_ERR (host->clk)) { -dev_err (&pdev->dev,"failed to get clock\n" ); err =PTR_ERR (host->clk); + +if (err != -EPROBE_DEFER ) +dev_err (&pdev->dev,"failed to get clock: %d\n" , err); +return err; }
The clock driver should be available at the time of host1x's probing on all Tegra's because it is one of essential core drivers that become available early during boot.
I guess you're making this change for T186, is it because the BPMP driver's probe getting deferred? If yes, won't it be possible to fix the defer of the clock driver instead of making such changes in the affected drivers?
Signature · 5 lines
Dmitry _______________________________________________ dri-devel mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/dri-devel
Quoted 17 lines · Dmitry Osipenko
04.06.2019 18:31, Thierry Reding пишет:
Quoted 16 lines
From: Thierry Reding <[email protected]>
When deferring probe, avoid logging a confusing error message. While at it, make the error message more informational.
Signed-off-by Thierry Reding <[email protected]>--- drivers/gpu/host1x/dev.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-)drivers/gpu/host1x/dev.c +0
-0@@ -247,8 +247,11 @@ static int host1x_probe(struct platform_device *pdev)
Quoted 14 lines
drivers/gpu/host1x/dev.c static int host1x_probe(struct platform_device *pdev) · line 256
Quoted 10 lines
drivers/gpu/host1x/dev.c +4
-1host->clk =devm_clk_get (&pdev->dev,NULL );if (IS_ERR (host->clk)) { -dev_err (&pdev->dev,"failed to get clock\n" ); err =PTR_ERR (host->clk); + +if (err != -EPROBE_DEFER ) +dev_err (&pdev->dev,"failed to get clock: %d\n" , err); +return err; }The clock driver should be available at the time of host1x's probing on all Tegra's because it is one of essential core drivers that become available early during boot.
That's the currently baked-in assumption. However, there can be any number of reasons for why the clocks may not show up as early as expected, as evidenced in the case of Tegra186.
Quoted 4 lines
I guess you're making this change for T186, is it because the BPMP driver's probe getting deferred? If yes, won't it be possible to fix the defer of the clock driver instead of making such changes in the affected drivers?
The reason why this is now happening on Tegra186 is because the BPMP is bound to an IOMMU to avoid getting faults from the new no-bypass policy that the ARM SMMU driver is implementing as of v5.2-rc1.
As a result of binding to an IOMMU, the first probe of the BPMP driver will get deferred, so any driver trying to request a clock after that and before BPMP gets probed successfully the next time, any clk_get() calls will fail with -EPROBE_DEFER.
This is a bit unfortunate, but like I said, these kinds of things can happen, and probe deferral was designed specifically to deal with that kind of situation so that we wouldn't have to rely on all of these built-in assumptions that occasionally break.
The driver also already handles deferred probe properly. The only thing that this patch really changes is to no longer consider -EPROBE_DEFER an error. It's in fact a pretty common situation in many drivers and should not warrant a kernel log message.
Thierry
_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel
05.06.2019 11:28, Thierry Reding пишет:
Quoted 56 lines · Dmitry Osipenko
Quoted 17 lines
04.06.2019 18:31, Thierry Reding пишет:
Quoted 16 lines
From: Thierry Reding <[email protected]>
When deferring probe, avoid logging a confusing error message. While at it, make the error message more informational.
Signed-off-by Thierry Reding <[email protected]>--- drivers/gpu/host1x/dev.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-)drivers/gpu/host1x/dev.c +0
-0@@ -247,8 +247,11 @@ static int host1x_probe(struct platform_device *pdev)Quoted 14 lines
drivers/gpu/host1x/dev.c static int host1x_probe(struct platform_device *pdev) · line 256
Quoted 10 lines
drivers/gpu/host1x/dev.c +4
-1host->clk =devm_clk_get (&pdev->dev,NULL );if (IS_ERR (host->clk)) { -dev_err (&pdev->dev,"failed to get clock\n" ); err =PTR_ERR (host->clk); + +if (err != -EPROBE_DEFER ) +dev_err (&pdev->dev,"failed to get clock: %d\n" , err); +return err; }The clock driver should be available at the time of host1x's probing on all Tegra's because it is one of essential core drivers that become available early during boot.
That's the currently baked-in assumption. However, there can be any number of reasons for why the clocks may not show up as early as expected, as evidenced in the case of Tegra186.
Quoted 4 lines
I guess you're making this change for T186, is it because the BPMP driver's probe getting deferred? If yes, won't it be possible to fix the defer of the clock driver instead of making such changes in the affected drivers?
The reason why this is now happening on Tegra186 is because the BPMP is bound to an IOMMU to avoid getting faults from the new no-bypass policy that the ARM SMMU driver is implementing as of v5.2-rc1.
As a result of binding to an IOMMU, the first probe of the BPMP driver will get deferred, so any driver trying to request a clock after that and before BPMP gets probed successfully the next time, any
clk_get()calls will fail with -EPROBE_DEFER.This is a bit unfortunate, but like I said, these kinds of things can happen, and probe deferral was designed specifically to deal with that kind of situation so that we wouldn't have to rely on all of these built-in assumptions that occasionally break.
Quoted 4 lines
The driver also already handles deferred probe properly. The only thing that this patch really changes is to no longer consider -EPROBE_DEFER an error. It's in fact a pretty common situation in many drivers and should not warrant a kernel log message.
You're trying to mask symptoms instead of curing the decease and it looks like the decease could be cured.
Won't something like this work for you?
From fbeabba5f1151e96edc38620db67593585558ca0 Mon Sep 17 00:00:00 2001
From: Dmitry Osipenko <[email protected]>
Date: Wed, 5 Jun 2019 14:02:00 +0300
Subject: [PATCH 1/2] iommu/arm-smmu: Move driver registration to subsys level
On some platforms there is a dependency on the IOMMU availability that comes up early during of the boot process. One example is NVIDIA Tegra186 which uses firmware, called BPMP, which manages lots of core functions like system clocks for example. That firmware driver require IOMMU functionality and hence the driver's probing is getting deferred because the ARM's SMMU driver is probed later, thus all the drivers that depend on the BPMP availability are also getting deferred on Tegra186. Let's move SMMU driver's registration to an earlier boot stage, allowing drivers like BPMP to probe successfully without the defer.
--- drivers/iommu/arm-smmu.c | 7 ++++++- 1 file changed, 6 insertions(+), 1 deletion(-)
drivers/iommu/arm-smmu.c +6 -1
@@ -2410,4 +2410,9 @@ static struct platform_driver arm_smmu_driver = { .probe = arm_smmu_device_probe, .shutdown = arm_smmu_device_shutdown, }; -builtin_platform_driver (arm_smmu_driver); + +static int __init arm_smmu_init (void ) +{ +return platform_driver_register (&arm_smmu_driver); +} +subsys_initcall (arm_smmu_init);
Signature · 17 lines
2.21.0 From 12ec90e22405b6d5574cbfcbd33b92736ad6bfe4 Mon Sep 17 00:00:00 2001 From: Dmitry Osipenko <[email protected]> Date: Wed, 5 Jun 2019 14:13:15 +0300 Subject: [PATCH 2/2] firmware/tegra: Move driver registration to subsys-sync level The BPMP driver depends on the ARM SMMU driver which is now getting probed from the subsys level, hence let's move the BPMP's probing to the subsys-sync level in order to avoid probe deferring of the BPMP's driver. Signed-off-by: Dmitry Osipenko <[email protected]> --- drivers/firmware/tegra/bpmp.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/firmware/tegra/bpmp.c +1 -1
@@ -883,4 +883,4 @@ static int __init tegra_bpmp_init(void) {return platform_driver_register (&tegra_bpmp_driver); } -core_initcall (tegra_bpmp_init); +subsys_initcall_sync (tegra_bpmp_init);
Signature · 7 lines
2.21.0 _______________________________________________ dri-devel mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/dri-devel
Quoted 57 lines · Dmitry Osipenko
05.06.2019 11:28, Thierry Reding пишет:
Quoted 56 lines · Dmitry Osipenko
Quoted 17 lines
04.06.2019 18:31, Thierry Reding пишет:
Quoted 16 lines
From: Thierry Reding <[email protected]>
When deferring probe, avoid logging a confusing error message. While at it, make the error message more informational.
Signed-off-by Thierry Reding <[email protected]>--- drivers/gpu/host1x/dev.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-)drivers/gpu/host1x/dev.c +0
-0@@ -247,8 +247,11 @@ static int host1x_probe(struct platform_device *pdev)Quoted 14 lines
drivers/gpu/host1x/dev.c static int host1x_probe(struct platform_device *pdev) · line 256
Quoted 10 lines
drivers/gpu/host1x/dev.c +4
-1host->clk =devm_clk_get (&pdev->dev,NULL );if (IS_ERR (host->clk)) { -dev_err (&pdev->dev,"failed to get clock\n" ); err =PTR_ERR (host->clk); + +if (err != -EPROBE_DEFER ) +dev_err (&pdev->dev,"failed to get clock: %d\n" , err); +return err; }The clock driver should be available at the time of host1x's probing on all Tegra's because it is one of essential core drivers that become available early during boot.
That's the currently baked-in assumption. However, there can be any number of reasons for why the clocks may not show up as early as expected, as evidenced in the case of Tegra186.
Quoted 4 lines
I guess you're making this change for T186, is it because the BPMP driver's probe getting deferred? If yes, won't it be possible to fix the defer of the clock driver instead of making such changes in the affected drivers?
The reason why this is now happening on Tegra186 is because the BPMP is bound to an IOMMU to avoid getting faults from the new no-bypass policy that the ARM SMMU driver is implementing as of v5.2-rc1.
As a result of binding to an IOMMU, the first probe of the BPMP driver will get deferred, so any driver trying to request a clock after that and before BPMP gets probed successfully the next time, any
clk_get()calls will fail with -EPROBE_DEFER.This is a bit unfortunate, but like I said, these kinds of things can happen, and probe deferral was designed specifically to deal with that kind of situation so that we wouldn't have to rely on all of these built-in assumptions that occasionally break.
Quoted 7 lines
Quoted 4 lines
The driver also already handles deferred probe properly. The only thing that this patch really changes is to no longer consider -EPROBE_DEFER an error. It's in fact a pretty common situation in many drivers and should not warrant a kernel log message.
You're trying to mask symptoms instead of curing the decease and it looks like the decease could be cured.
There's nothing here to cure. -EPROBE_DEFER was designed specifically to avoid having to play these kinds of games with initcall levels.
What this patch tries to do is just to avoid printing an error message for something that is not an error. -EPROBE_DEFER is totally expected to happen, it's normal, it's not something that we should bother users with because things end up sorting themselves out in the end.
Quoted 1 line
Won't something like this work for you?
I'm sure we could find a number of ways to fix this. But there's no need to fix this because it's not broken. What is broken is that we output an error message when this happens and make an elephant out of a fly.
Thierry
Quoted 68 lines
From
fbeabba5f1151e96edc38620db67593585558ca0Mon Sep 17 00:00:00 2001
From: Dmitry Osipenko <[email protected]>
Date: Wed, 5 Jun 2019 14:02:00 +0300
Subject: [PATCH 1/2] iommu/arm-smmu: Move driver registration to subsys levelOn some platforms there is a dependency on the IOMMU availability that comes up early during of the boot process. One example is NVIDIA Tegra186 which uses firmware, called BPMP, which manages lots of core functions like system clocks for example. That firmware driver require IOMMU functionality and hence the driver's probing is getting deferred because the ARM's SMMU driver is probed later, thus all the drivers that depend on the BPMP availability are also getting deferred on Tegra186. Let's move SMMU driver's registration to an earlier boot stage, allowing drivers like BPMP to probe successfully without the defer.
Signed-off-by Dmitry Osipenko <[email protected]>--- drivers/iommu/arm-smmu.c | 7 ++++++- 1 file changed, 6 insertions(+), 1 deletion(-)drivers/iommu/arm-smmu.c +6
-1@@ -2410,4 +2410,9 @@ static struct platform_driver arm_smmu_driver = { .probe = arm_smmu_device_probe, .shutdown = arm_smmu_device_shutdown, }; -builtin_platform_driver (arm_smmu_driver); + +static int __init arm_smmu_init (void ) +{ +return platform_driver_register (&arm_smmu_driver); +} +subsys_initcall (arm_smmu_init);Signature · 17 lines
2.21.0 From 12ec90e22405b6d5574cbfcbd33b92736ad6bfe4 Mon Sep 17 00:00:00 2001 From: Dmitry Osipenko <[email protected]> Date: Wed, 5 Jun 2019 14:13:15 +0300 Subject: [PATCH 2/2] firmware/tegra: Move driver registration to subsys-sync level The BPMP driver depends on the ARM SMMU driver which is now getting probed from the subsys level, hence let's move the BPMP's probing to the subsys-sync level in order to avoid probe deferring of the BPMP's driver. Signed-off-by: Dmitry Osipenko <[email protected]> --- drivers/firmware/tegra/bpmp.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)drivers/firmware/tegra/bpmp.c +1
-1@@ -883,4 +883,4 @@ static int __init tegra_bpmp_init(void) {return platform_driver_register (&tegra_bpmp_driver); } -core_initcall (tegra_bpmp_init); +subsys_initcall_sync (tegra_bpmp_init);Signature · 1 line
2.21.0
_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel
05.06.2019 15:32, Thierry Reding пишет:
Quoted 74 lines · Dmitry Osipenko
Quoted 57 lines
05.06.2019 11:28, Thierry Reding пишет:
Quoted 56 lines · Dmitry Osipenko
Quoted 17 lines
04.06.2019 18:31, Thierry Reding пишет:
Quoted 16 lines
From: Thierry Reding <[email protected]>
When deferring probe, avoid logging a confusing error message. While at it, make the error message more informational.
Signed-off-by Thierry Reding <[email protected]>--- drivers/gpu/host1x/dev.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-)drivers/gpu/host1x/dev.c +0
-0@@ -247,8 +247,11 @@ static int host1x_probe(struct platform_device *pdev)Quoted 14 lines
drivers/gpu/host1x/dev.c static int host1x_probe(struct platform_device *pdev) · line 256
Quoted 10 lines
drivers/gpu/host1x/dev.c +4
-1host->clk =devm_clk_get (&pdev->dev,NULL );if (IS_ERR (host->clk)) { -dev_err (&pdev->dev,"failed to get clock\n" ); err =PTR_ERR (host->clk); + +if (err != -EPROBE_DEFER ) +dev_err (&pdev->dev,"failed to get clock: %d\n" , err); +return err; }The clock driver should be available at the time of host1x's probing on all Tegra's because it is one of essential core drivers that become available early during boot.
That's the currently baked-in assumption. However, there can be any number of reasons for why the clocks may not show up as early as expected, as evidenced in the case of Tegra186.
Quoted 4 lines
I guess you're making this change for T186, is it because the BPMP driver's probe getting deferred? If yes, won't it be possible to fix the defer of the clock driver instead of making such changes in the affected drivers?
The reason why this is now happening on Tegra186 is because the BPMP is bound to an IOMMU to avoid getting faults from the new no-bypass policy that the ARM SMMU driver is implementing as of v5.2-rc1.
As a result of binding to an IOMMU, the first probe of the BPMP driver will get deferred, so any driver trying to request a clock after that and before BPMP gets probed successfully the next time, any
clk_get()calls will fail with -EPROBE_DEFER.This is a bit unfortunate, but like I said, these kinds of things can happen, and probe deferral was designed specifically to deal with that kind of situation so that we wouldn't have to rely on all of these built-in assumptions that occasionally break.
Quoted 7 lines
Quoted 4 lines
The driver also already handles deferred probe properly. The only thing that this patch really changes is to no longer consider -EPROBE_DEFER an error. It's in fact a pretty common situation in many drivers and should not warrant a kernel log message.
You're trying to mask symptoms instead of curing the decease and it looks like the decease could be cured.
There's nothing here to cure. -EPROBE_DEFER was designed specifically to avoid having to play these kinds of games with initcall levels.
What this patch tries to do is just to avoid printing an error message for something that is not an error. -EPROBE_DEFER is totally expected to happen, it's normal, it's not something that we should bother users with because things end up sorting themselves out in the end.
Quoted 5 lines
Quoted 1 line
Won't something like this work for you?
I'm sure we could find a number of ways to fix this. But there's no need to fix this because it's not broken. What is broken is that we output an error message when this happens and make an elephant out of a fly.
Sure, this is absolutely not critical and deferred probe is doing its job. But don't you agree that it's better to fix the root of the annoyance once and for all?
Would be awesome if you could give a whirl to the patches that I'm suggesting.
_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel
Quoted 77 lines · Dmitry Osipenko
05.06.2019 15:32, Thierry Reding пишет:
Quoted 76 lines · Dmitry Osipenko
Quoted 57 lines
05.06.2019 11:28, Thierry Reding пишет:
Quoted 56 lines · Dmitry Osipenko
Quoted 17 lines
04.06.2019 18:31, Thierry Reding пишет:
Quoted 16 lines
From: Thierry Reding <[email protected]>
When deferring probe, avoid logging a confusing error message. While at it, make the error message more informational.
Signed-off-by Thierry Reding <[email protected]>--- drivers/gpu/host1x/dev.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-)drivers/gpu/host1x/dev.c +0
-0@@ -247,8 +247,11 @@ static int host1x_probe(struct platform_device *pdev)Quoted 14 lines
drivers/gpu/host1x/dev.c static int host1x_probe(struct platform_device *pdev) · line 256
Quoted 10 lines
drivers/gpu/host1x/dev.c +4
-1host->clk =devm_clk_get (&pdev->dev,NULL );if (IS_ERR (host->clk)) { -dev_err (&pdev->dev,"failed to get clock\n" ); err =PTR_ERR (host->clk); + +if (err != -EPROBE_DEFER ) +dev_err (&pdev->dev,"failed to get clock: %d\n" , err); +return err; }The clock driver should be available at the time of host1x's probing on all Tegra's because it is one of essential core drivers that become available early during boot.
That's the currently baked-in assumption. However, there can be any number of reasons for why the clocks may not show up as early as expected, as evidenced in the case of Tegra186.
Quoted 4 lines
I guess you're making this change for T186, is it because the BPMP driver's probe getting deferred? If yes, won't it be possible to fix the defer of the clock driver instead of making such changes in the affected drivers?
The reason why this is now happening on Tegra186 is because the BPMP is bound to an IOMMU to avoid getting faults from the new no-bypass policy that the ARM SMMU driver is implementing as of v5.2-rc1.
As a result of binding to an IOMMU, the first probe of the BPMP driver will get deferred, so any driver trying to request a clock after that and before BPMP gets probed successfully the next time, any
clk_get()calls will fail with -EPROBE_DEFER.This is a bit unfortunate, but like I said, these kinds of things can happen, and probe deferral was designed specifically to deal with that kind of situation so that we wouldn't have to rely on all of these built-in assumptions that occasionally break.
Quoted 7 lines
Quoted 4 lines
The driver also already handles deferred probe properly. The only thing that this patch really changes is to no longer consider -EPROBE_DEFER an error. It's in fact a pretty common situation in many drivers and should not warrant a kernel log message.
You're trying to mask symptoms instead of curing the decease and it looks like the decease could be cured.
There's nothing here to cure. -EPROBE_DEFER was designed specifically to avoid having to play these kinds of games with initcall levels.
What this patch tries to do is just to avoid printing an error message for something that is not an error. -EPROBE_DEFER is totally expected to happen, it's normal, it's not something that we should bother users with because things end up sorting themselves out in the end.
Quoted 1 line
Won't something like this work for you?
Quoted 7 lines
Quoted 3 lines
I'm sure we could find a number of ways to fix this. But there's no need to fix this because it's not broken. What is broken is that we output an error message when this happens and make an elephant out of a fly.
Sure, this is absolutely not critical and deferred probe is doing its job. But don't you agree that it's better to fix the root of the annoyance once and for all?
From my point of view deferred probe is the once and for all fix. Back before we had deferred probe, doing these kinds of initcall reordering tricks was fairly common and while such a change may fix one setup, it often ended up breaking others.
Sorry, this is a lesson that we already learned a couple of years ago, no need to rehash it.
Thierry
_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel
05.06.2019 16:04, Thierry Reding пишет:
Quoted 86 lines · Dmitry Osipenko
Quoted 77 lines
05.06.2019 15:32, Thierry Reding пишет:
Quoted 76 lines · Dmitry Osipenko
Quoted 57 lines
05.06.2019 11:28, Thierry Reding пишет:
Quoted 56 lines · Dmitry Osipenko
Quoted 17 lines
04.06.2019 18:31, Thierry Reding пишет:
Quoted 16 lines
From: Thierry Reding <[email protected]>
When deferring probe, avoid logging a confusing error message. While at it, make the error message more informational.
Signed-off-by Thierry Reding <[email protected]>--- drivers/gpu/host1x/dev.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-)drivers/gpu/host1x/dev.c +0
-0@@ -247,8 +247,11 @@ static int host1x_probe(struct platform_device *pdev)Quoted 14 lines
drivers/gpu/host1x/dev.c static int host1x_probe(struct platform_device *pdev) · line 256
Quoted 10 lines
drivers/gpu/host1x/dev.c +4
-1host->clk =devm_clk_get (&pdev->dev,NULL );if (IS_ERR (host->clk)) { -dev_err (&pdev->dev,"failed to get clock\n" ); err =PTR_ERR (host->clk); + +if (err != -EPROBE_DEFER ) +dev_err (&pdev->dev,"failed to get clock: %d\n" , err); +return err; }The clock driver should be available at the time of host1x's probing on all Tegra's because it is one of essential core drivers that become available early during boot.
That's the currently baked-in assumption. However, there can be any number of reasons for why the clocks may not show up as early as expected, as evidenced in the case of Tegra186.
Quoted 4 lines
I guess you're making this change for T186, is it because the BPMP driver's probe getting deferred? If yes, won't it be possible to fix the defer of the clock driver instead of making such changes in the affected drivers?
The reason why this is now happening on Tegra186 is because the BPMP is bound to an IOMMU to avoid getting faults from the new no-bypass policy that the ARM SMMU driver is implementing as of v5.2-rc1.
As a result of binding to an IOMMU, the first probe of the BPMP driver will get deferred, so any driver trying to request a clock after that and before BPMP gets probed successfully the next time, any
clk_get()calls will fail with -EPROBE_DEFER.This is a bit unfortunate, but like I said, these kinds of things can happen, and probe deferral was designed specifically to deal with that kind of situation so that we wouldn't have to rely on all of these built-in assumptions that occasionally break.
Quoted 7 lines
Quoted 4 lines
The driver also already handles deferred probe properly. The only thing that this patch really changes is to no longer consider -EPROBE_DEFER an error. It's in fact a pretty common situation in many drivers and should not warrant a kernel log message.
You're trying to mask symptoms instead of curing the decease and it looks like the decease could be cured.
There's nothing here to cure. -EPROBE_DEFER was designed specifically to avoid having to play these kinds of games with initcall levels.
What this patch tries to do is just to avoid printing an error message for something that is not an error. -EPROBE_DEFER is totally expected to happen, it's normal, it's not something that we should bother users with because things end up sorting themselves out in the end.
Quoted 1 line
Won't something like this work for you?
Quoted 7 lines
Quoted 3 lines
I'm sure we could find a number of ways to fix this. But there's no need to fix this because it's not broken. What is broken is that we output an error message when this happens and make an elephant out of a fly.
Sure, this is absolutely not critical and deferred probe is doing its job. But don't you agree that it's better to fix the root of the annoyance once and for all?
Quoted 7 lines
From my point of view deferred probe is the once and for all fix. Back before we had deferred probe, doing these kinds of initcall reordering tricks was fairly common and while such a change may fix one setup, it often ended up breaking others.
Sorry, this is a lesson that we already learned a couple of years ago, no need to rehash it.
We had a success story in similar case of pinctrl vs GPIO driver probe order, reordering the probe order fixed a lot of deferred probings and improved boot up time a tad. Oh, that reminded me that the Stefan's patch was never merged [0] :(
I'm trying to help, but seems not much I could do since you're in oppose ;)
[0] https://patchwork.ozlabs.org/patch/949776/
_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel
04.06.2019 18:31, Thierry Reding пишет:
Quoted 16 lines
From: Thierry Reding <[email protected]>
When deferring probe, avoid logging a confusing error message. While at it, make the error message more informational.
Signed-off-by Thierry Reding <[email protected]>--- drivers/gpu/host1x/dev.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-)drivers/gpu/host1x/dev.c +0
-0@@ -247,8 +247,11 @@ static int host1x_probe(struct platform_device *pdev)
drivers/gpu/host1x/dev.c static int host1x_probe(struct platform_device *pdev) · line 256
Quoted 10 lines
drivers/gpu/host1x/dev.c +4
-1host->clk =devm_clk_get (&pdev->dev,NULL );if (IS_ERR (host->clk)) { -dev_err (&pdev->dev,"failed to get clock\n" ); err =PTR_ERR (host->clk); + +if (err != -EPROBE_DEFER ) +dev_err (&pdev->dev,"failed to get clock: %d\n" , err); +return err; }
The clock driver should be available at the time of host1x's probing on all Tegra's because it is one of essential core drivers that become available early during boot.
I guess you're making this change for T186, is it because the BPMP driver's probe getting deferred? If yes, won't it be possible to fix the defer of the clock driver instead of making such changes in all of the affected drivers?
[it appeared to me that first email got dropped by gmail, so I'm re-sending it second time just in a case]
Signature · 5 lines
Dmitry _______________________________________________ dri-devel mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/dri-devel
Quoted 15 lines · Thierry Reding
From: Thierry Reding <[email protected]>
When deferring probe, avoid logging a confusing error message. While at it, make the error message more informational.
Signed-off-by Thierry Reding <[email protected]>--- drivers/gpu/host1x/dev.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-)drivers/gpu/host1x/dev.c +0
-0
drivers/gpu/host1x/dev.c static int host1x_probe(struct platform_device *pdev) · line 253
Quoted 8 lines
drivers/gpu/host1x/dev.c +3
-1@@ -247,8 +247,11 @@ static int host1x_probe(struct platform_device *pdev) host->clk =devm_clk_get (&pdev->dev,NULL );if (IS_ERR (host->clk)) { -dev_err (&pdev->dev,"failed to get clock\n" ); err =PTR_ERR (host->clk); + +if (err != -EPROBE_DEFER ) +dev_err (&pdev->dev,"failed to get clock: %d\n" , err);
Quoted 11 lines
drivers/gpu/host1x/dev.c +1
-0+return err; }Signature · 6 lines
2.21.0 _______________________________________________ dri-devel mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/dri-devel
Signature · 7 lines
Daniel Vetter Software Engineer, Intel Corporation http://blog.ffwll.ch _______________________________________________ dri-devel mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/dri-devel