Skip to contentSkip to conversation

dt/sparc: Introduce the use of the managed version of kzalloc

PATCH In review Himangi Saraogi 3 participants 6 messages May 22, 2014 – May 23, 2014

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 dt/sparc: Introduce the use of the managed version of kzalloc

This patch moves data allocated using kzalloc to managed data allocated using devm_kzalloc and cleans now unnecessary kfrees in probe and remove functions. Also the unnecessary labels are done away with.

The following Coccinelle semantic patch was used for making the change:

@platform@
identifier p, probefn, removefn;
@@
struct platform_driver p = {
.probe = probefn,
.remove = removefn,
};

@prb@
identifier platform.probefn, pdev;
expression e, e1, e2;
@@
probefn(struct platform_device *pdev, ...) {
<+...
- e = kzalloc(e1, e2)
+ e = devm_kzalloc(&pdev->dev, e1, e2)
...
?-kfree(e);
+kfree(DROPME,e);
...+>
}

@rem depends on prb@
identifier platform.removefn;
expression e;
@@
removefn(...) {
<...
- kfree(e);
...>
}

@label1@
identifier platform.probefn, l1, l2;
expression e;
@@
probefn(...) {
<+...
l1:kfree(DROPME, e);
l2:
...+>
}

@rem_label depends on label1@
identifier platform.probefn,label1.l1,label1.l2;
expression e;
@@
probefn(...) {
<+...
-goto l1;
+goto l2;
...
-l1:kfree(DROPME, e);
...+>
}

---
Not compiled due to incompatible architecture.

 drivers/input/misc/sparcspkr.c | 29 +++++++++--------------------
 1 file changed, 9 insertions(+), 20 deletions(-)
drivers/input/misc/sparcspkr.c +9 -20
index 65fd315..da9cdc6 100644
@@ -187,29 +187,28 @@ static int bbc_beep_probe(struct platform_device *op)
 	struct sparcspkr_state *state;
 	struct bbc_beep_info *info;
 	struct device_node *dp;
-	int err = -ENOMEM;
+	int err;
 
-	state = kzalloc(sizeof(*state), GFP_KERNEL);
+	state = devm_kzalloc(&op->dev, sizeof(*state), GFP_KERNEL);
 	if (!state)
-		goto out_err;
+		return -ENOMEM;
 
 	state->name = "Sparc BBC Speaker";
 	state->event = bbc_spkr_event;
 	spin_lock_init(&state->lock);
 
 	dp = of_find_node_by_path("/");
-	err = -ENODEV;
 	if (!dp)
-		goto out_free;
+		return -ENODEV;
 
 	info = &state->u.bbc;
 	info->clock_freq = of_getintprop_default(dp, "clock-frequency", 0);
 	if (!info->clock_freq)
-		goto out_free;
+		return -ENODEV;
 
 	info->regs = of_ioremap(&op->resource[0], 0, 6, "bbc beep");
 	if (!info->regs)
-		goto out_free;
+		return -ENODEV;
 
 	platform_set_drvdata(op, state);
 
@@ -222,9 +221,6 @@ static int bbc_beep_probe(struct platform_device *op)
 out_clear_drvdata:
 	of_iounmap(&op->resource[0], info->regs, 6);
 
-out_free:
-	kfree(state);
-out_err:
 	return err;
 }
 
@@ -241,8 +237,6 @@ static int bbc_remove(struct platform_device *op)
 
 	of_iounmap(&op->resource[0], info->regs, 6);
 
-	kfree(state);
-
 	return 0;
 }
 
@@ -271,9 +265,9 @@ static int grover_beep_probe(struct platform_device *op)
 	struct grover_beep_info *info;
 	int err = -ENOMEM;
 
-	state = kzalloc(sizeof(*state), GFP_KERNEL);
+	state = devm_kzalloc(&op->dev, sizeof(*state), GFP_KERNEL);
 	if (!state)
-		goto out_err;
+		return -ENOMEM;
 
 	state->name = "Sparc Grover Speaker";
 	state->event = grover_spkr_event;
@@ -282,7 +276,7 @@ static int grover_beep_probe(struct platform_device *op)
 	info = &state->u.grover;
 	info->freq_regs = of_ioremap(&op->resource[2], 0, 2, "grover beep freq");
 	if (!info->freq_regs)
-		goto out_free;
+		return -ENOMEM;
 
 	info->enable_reg = of_ioremap(&op->resource[3], 0, 1, "grover beep enable");
 	if (!info->enable_reg)
@@ -301,9 +295,6 @@ out_clear_drvdata:
 
 out_unmap_freq_regs:
 	of_iounmap(&op->resource[2], info->freq_regs, 2);
-out_free:
-	kfree(state);
-out_err:
 	return err;
 }
 
@@ -321,8 +312,6 @@ static int grover_remove(struct platform_device *op)
 	of_iounmap(&op->resource[3], info->enable_reg, 1);
 	of_iounmap(&op->resource[2], info->freq_regs, 2);
 
-	kfree(state);
-
 	return 0;
 }
 
Signature · 1 line
1.9.1
maintainer
#1 lore raw

Hi Himangi,

Quoted 3 lines · Himangi Saraogi

This patch moves data allocated using kzalloc to managed data allocated using devm_kzalloc and cleans now unnecessary kfrees in probe and remove functions. Also the unnecessary labels are done away with.

Please no more partial devm conversions. I am not interested in patches that change only 1 resource to be managed and leaving the rest as is.

If you want to do devm conversion please do it if you convert all (or at least majority) of resource handling to managed resources.

...

Quoted 3 lines

---
Not compiled due to incompatible architecture.

There are cross-compilers that can be used to work around such issues.

Thanks.

Signature · 1 line
Dmitry
#2 lore raw
Quoted 7 lines · Dmitry Torokhov

Hi Himangi,

Quoted 3 lines · Himangi Saraogi

This patch moves data allocated using kzalloc to managed data allocated using devm_kzalloc and cleans now unnecessary kfrees in probe and remove functions. Also the unnecessary labels are done away with.

Quoted 5 lines

Please no more partial devm conversions. I am not interested in patches that change only 1 resource to be managed and leaving the rest as is.

If you want to do devm conversion please do it if you convert all (or at least majority) of resource handling to managed resources.

Sorry, I don't think we are clear on what is wanted. Himangi's patches convert everything where a devm function is already available (to our knowledge). Do you prefer that new devm functions be introduced? On the one hand, that is perhaps useful from a safety point of view. On the other hand, it could lead to a huge number of rarely used devm functions that no one knows about and to a lot of code diversity. So I don't know what is the good solution.

thanks,
julia

maintainer
#3 lore raw

Hi Julia,

Quoted 17 lines · Julia Lawall
Quoted 7 lines · Dmitry Torokhov

Hi Himangi,

Quoted 3 lines · Himangi Saraogi

This patch moves data allocated using kzalloc to managed data allocated using devm_kzalloc and cleans now unnecessary kfrees in probe and remove functions. Also the unnecessary labels are done away with.

Quoted 5 lines

Please no more partial devm conversions. I am not interested in patches that change only 1 resource to be managed and leaving the rest as is.

If you want to do devm conversion please do it if you convert all (or at least majority) of resource handling to managed resources.

Quoted 7 lines

Sorry, I don't think we are clear on what is wanted. Himangi's patches convert everything where a devm function is already available (to our knowledge). Do you prefer that new devm functions be introduced? On the one hand, that is perhaps useful from a safety point of view. On the other hand, it could lead to a huge number of rarely used devm functions that no one knows about and to a lot of code diversity. So I don't know what is the good solution.

I strongly dislike changes that mix managed and non-managed resources in the same driver as it only leads to confusion: should this resource be freed or will it be freed automatically.

I am applying patches that convert all (or almost all) resources in a driver into managed ones, but if that is not [currently] possible then those drivers are probably better left alone (maybe after additional audit).

Thanks.

Signature · 1 line
Dmitry
#4 lore raw
Quoted 28 lines · Dmitry Torokhov

Hi Julia,

Quoted 17 lines · Julia Lawall
Quoted 7 lines · Dmitry Torokhov

Hi Himangi,

Quoted 3 lines · Himangi Saraogi

This patch moves data allocated using kzalloc to managed data allocated using devm_kzalloc and cleans now unnecessary kfrees in probe and remove functions. Also the unnecessary labels are done away with.

Quoted 5 lines

Please no more partial devm conversions. I am not interested in patches that change only 1 resource to be managed and leaving the rest as is.

If you want to do devm conversion please do it if you convert all (or at least majority) of resource handling to managed resources.

Quoted 7 lines

Sorry, I don't think we are clear on what is wanted. Himangi's patches convert everything where a devm function is already available (to our knowledge). Do you prefer that new devm functions be introduced? On the one hand, that is perhaps useful from a safety point of view. On the other hand, it could lead to a huge number of rarely used devm functions that no one knows about and to a lot of code diversity. So I don't know what is the good solution.

Quoted 8 lines

I strongly dislike changes that mix managed and non-managed resources in the same driver as it only leads to confusion: should this resource be freed or will it be freed automatically.

I am applying patches that convert all (or almost all) resources in a driver into managed ones, but if that is not [currently] possible then those drivers are probably better left alone (maybe after additional audit).

OK, we will try to take that into account. Are the new managed functoins added to the devres docuentation?

thanks,
julia

#5 lore raw
In reply to Julia Lawall
Quoted 17 lines · Julia Lawall
Quoted 7 lines · Dmitry Torokhov

Hi Himangi,

Quoted 3 lines · Himangi Saraogi

This patch moves data allocated using kzalloc to managed data allocated using devm_kzalloc and cleans now unnecessary kfrees in probe and remove functions. Also the unnecessary labels are done away with.

Quoted 5 lines

Please no more partial devm conversions. I am not interested in patches that change only 1 resource to be managed and leaving the rest as is.

If you want to do devm conversion please do it if you convert all (or at least majority) of resource handling to managed resources.

Quoted 7 lines

Sorry, I don't think we are clear on what is wanted. Himangi's patches convert everything where a devm function is already available (to our knowledge). Do you prefer that new devm functions be introduced? On the one hand, that is perhaps useful from a safety point of view. On the other hand, it could lead to a huge number of rarely used devm functions that no one knows about and to a lot of code diversity. So I don't know what is the good solution.

I guess that the concrete problem is of_ioremap? Since that is a pretty common function, I guess it could be reasonable to introduce a devm function for it as well. Or maybe one already exists and we overlooked it.

julia