linux-pm.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Hans de Goede <hansg@kernel.org>
To: Sebastian Reichel <sebastian.reichel@collabora.com>,
	Sebastian Reichel <sre@kernel.org>,
	Mark Brown <broonie@kernel.org>,
	Linus Walleij <linus.walleij@linaro.org>
Cc: "Liam Girdwood" <lgirdwood@gmail.com>,
	"Vinod Koul" <vkoul@kernel.org>,
	"Kishon Vijay Abraham I" <kishon@kernel.org>,
	"Chen-Yu Tsai" <wens@csie.org>,
	"Jernej Skrabec" <jernej.skrabec@gmail.com>,
	"Samuel Holland" <samuel@sholland.org>,
	"Matti Vaittinen" <mazziesaccount@gmail.com>,
	"Pali Rohár" <pali@kernel.org>,
	"Krzysztof Kozlowski" <krzk@kernel.org>,
	"AngeloGioacchino Del Regno"
	<angelogioacchino.delregno@collabora.com>,
	linux-pm@vger.kernel.org, linux-kernel@vger.kernel.org
Subject: Re: [PATCH v2 3/5] power: supply: core: battery-info: fully switch to fwnode
Date: Sun, 8 Jun 2025 22:16:06 +0200	[thread overview]
Message-ID: <c76fe3f8-8884-4769-af69-a9a527d2a202@kernel.org> (raw)
In-Reply-To: <68792092-4689-4e3e-8dbb-9157889f1a3d@kernel.org>

[-- Attachment #1: Type: text/plain, Size: 9521 bytes --]

Hi,

On 8-Jun-25 9:57 PM, Hans de Goede wrote:
> Hi Sebastian,
> On 30-Apr-25 12:54 AM, Sebastian Reichel wrote:
>> Also use fwnode based parsing for "ocv-capacity-celsius" and
>> "resistance-temp-table", so that any DT specific bits are
>> removed from the power-supply core.
>>
>> Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
>> Signed-off-by: Sebastian Reichel <sebastian.reichel@collabora.com>
> 
> I have been testing this converting the ug3105 driver to
> use power_supply_batinfo_ocv2cap(), replacing the hardcoded
> ocv -> capacity table in that driver.
> 
> While testing I hit a bug and while looking closer at this
> patch it needs more work on top of fixing that bug.
> 
> See comments inline, also I've attached 3 fixup patches
> which can be squashed into this patch to address the remarks.
> 
>> ---
>>  drivers/power/supply/power_supply_core.c | 109 ++++++++++++++++++-------------
>>  1 file changed, 63 insertions(+), 46 deletions(-)
>>
>> diff --git a/drivers/power/supply/power_supply_core.c b/drivers/power/supply/power_supply_core.c
>> index 89947f1fe610d8a75756e1e4e5339b06349f9ab8..a8d1fe66e2486a833ccaa3ed77b861c6e52c5760 100644
>> --- a/drivers/power/supply/power_supply_core.c
>> +++ b/drivers/power/supply/power_supply_core.c
>> @@ -585,32 +585,19 @@ int power_supply_get_battery_info(struct power_supply *psy,
>>  {
>>  	struct power_supply_resistance_temp_table *resist_table;
>>  	struct power_supply_battery_info *info;
>> -	struct device_node *battery_np = NULL;
>> -	struct fwnode_reference_args args;
>> -	struct fwnode_handle *fwnode = NULL;
>> +	struct fwnode_handle *srcnode, *fwnode;
>>  	const char *value;
>> -	int err, len, index;
>> -	const __be32 *list;
>> +	int err, len, index, proplen;
>> +	u32 *propdata;
> 
> propname which is also a local-variable for a temporary
> malloc-ed buffer uses __free(kfree) instead of explicit
> kfree() calls IMHO it would be good to do that here too.
> 
> This requires declaring it inside the
> "for (index = 0; index < len; index++)" loop like how
> this is done for propname, so that it gets freed on
> every loop iteration.
> 
> 
>>  	u32 min_max[2];
>>  
>> -	if (psy->dev.of_node) {
>> -		battery_np = of_parse_phandle(psy->dev.of_node, "monitored-battery", 0);
>> -		if (!battery_np)
>> -			return -ENODEV;
>> +	srcnode = dev_fwnode(&psy->dev);
>> +	if (!srcnode && psy->dev.parent)
>> +		srcnode = dev_fwnode(psy->dev.parent);
>>  
>> -		fwnode = fwnode_handle_get(of_fwnode_handle(battery_np));
>> -	} else if (psy->dev.parent) {
>> -		err = fwnode_property_get_reference_args(
>> -					dev_fwnode(psy->dev.parent),
>> -					"monitored-battery", NULL, 0, 0, &args);
>> -		if (err)
>> -			return err;
>> -
>> -		fwnode = args.fwnode;
>> -	}
>> -
>> -	if (!fwnode)
>> -		return -ENOENT;
>> +	fwnode = fwnode_find_reference(srcnode, "monitored-battery", 0);
>> +	if (IS_ERR(fwnode))
>> +		return PTR_ERR(fwnode);
>>  
>>  	err = fwnode_property_read_string(fwnode, "compatible", &value);
>>  	if (err)
>> @@ -740,15 +727,7 @@ int power_supply_get_battery_info(struct power_supply *psy,
>>  		info->temp_max = min_max[1];
>>  	}
>>  
>> -	/*
>> -	 * The below code uses raw of-data parsing to parse
>> -	 * /schemas/types.yaml#/definitions/uint32-matrix
>> -	 * data, so for now this is only support with of.
>> -	 */
>> -	if (!battery_np)
>> -		goto out_ret_pointer;
>> -
>> -	len = of_property_count_u32_elems(battery_np, "ocv-capacity-celsius");
>> +	len = fwnode_property_count_u32(fwnode, "ocv-capacity-celsius");
>>  	if (len < 0 && len != -EINVAL) {
>>  		err = len;
>>  		goto out_put_node;
>> @@ -757,13 +736,13 @@ int power_supply_get_battery_info(struct power_supply *psy,
>>  		err = -EINVAL;
>>  		goto out_put_node;
>>  	} else if (len > 0) {
>> -		of_property_read_u32_array(battery_np, "ocv-capacity-celsius",
>> +		fwnode_property_read_u32_array(fwnode, "ocv-capacity-celsius",
>>  					   info->ocv_temp, len);
>>  	}
>>  
>>  	for (index = 0; index < len; index++) {
>>  		struct power_supply_battery_ocv_table *table;
>> -		int i, tab_len, size;
>> +		int i, tab_len;
>>  
>>  		char *propname __free(kfree) = kasprintf(GFP_KERNEL, "ocv-capacity-table-%d",
>>  							 index);
>> @@ -772,60 +751,98 @@ int power_supply_get_battery_info(struct power_supply *psy,
>>  			err = -ENOMEM;
>>  			goto out_put_node;
>>  		}
>> -		list = of_get_property(battery_np, propname, &size);
>> -		if (!list || !size) {
>> +		proplen = fwnode_property_count_u32(fwnode, propname);
>> +		if (proplen < 0 || proplen % 2 != 0) {
>>  			dev_err(&psy->dev, "failed to get %s\n", propname);
>>  			power_supply_put_battery_info(psy, info);
>>  			err = -EINVAL;
>>  			goto out_put_node;
>>  		}
>> +		propdata = kcalloc(proplen, sizeof(*propdata), GFP_KERNEL);
> 
> As mentioned above I suggest to use the following here instead:
> 
> 		u32 *propdata __free(kfree) = kcalloc(proplen, sizeof(*propdata), GFP_KERNEL);
> 		
> 
>> +		if (!propdata) {
>> +			kfree(propname);
> 
> propname must NOT be free-ed here since it is marked __free(kfree),
> freeing it here will cause a double-free bug.
> 
>> +			power_supply_put_battery_info(psy, info);
>> +			err = -EINVAL;
>> +			goto out_put_node;
>> +		}
>> +		err = fwnode_property_read_u32_array(fwnode, propname, propdata, proplen);
>> +		if (err < 0) {
>> +			dev_err(&psy->dev, "failed to get %s\n", propname);
>> +			kfree(propname);
> 
> same as above.
> 
>> +			kfree(propdata);
> 
> with the suggested "u32 *propdata __free(kfree)" this can and must be dropped.
> 
>> +			power_supply_put_battery_info(psy, info);
>> +			goto out_put_node;
>> +		}
>>  
>> -		tab_len = size / (2 * sizeof(__be32));
>> +		tab_len = proplen / 2;
>>  		info->ocv_table_size[index] = tab_len;
>>  
>>  		info->ocv_table[index] = table =
>>  			devm_kcalloc(&psy->dev, tab_len, sizeof(*table), GFP_KERNEL);
>>  		if (!info->ocv_table[index]) {
>> +			kfree(propdata);
> 
> with the suggested "u32 *propdata __free(kfree)" this can and must be dropped.
> 
>>  			power_supply_put_battery_info(psy, info);
>>  			err = -ENOMEM;
>>  			goto out_put_node;
>>  		}
>>  
>>  		for (i = 0; i < tab_len; i++) {
>> -			table[i].ocv = be32_to_cpu(*list);
>> -			list++;
>> -			table[i].capacity = be32_to_cpu(*list);
>> -			list++;
>> +			table[i].ocv = propdata[i*2];
>> +			table[i].capacity = propdata[i*2+1];
>>  		}
>> +
>> +		kfree(propdata);
> 
> with the suggested "u32 *propdata __free(kfree)" this can and must be dropped.
> 
>>  	}
>>  
>> -	list = of_get_property(battery_np, "resistance-temp-table", &len);
>> -	if (!list || !len)
>> +	proplen = fwnode_property_count_u32(fwnode, "resistance-temp-table");
> 
> This will return -EINVAL when the property does not exist, making
> power_supply_get_battery_info() always fail when there is no
> "resistance-temp-table" in the battery fwnode. See the attached fixup
> patch for a suggested fix.
> 
>> +	if (proplen < 0 || proplen % 2 != 0) {
>> +		power_supply_put_battery_info(psy, info);
>> +		err = -ENOMEM;
> 
> -ENOMEM is the wrong error code here.
> 
>>  		goto out_ret_pointer;
> 
> This should be "goto out_put_node" since this is an error path.
> 
>> +	} else if (proplen == 0) {
>> +		goto out_ret_pointer;
>> +	}
>>  
>> -	info->resist_table_size = len / (2 * sizeof(__be32));
>> +	propdata = kcalloc(proplen, sizeof(*propdata), GFP_KERNEL);
> 
> As mentioned above I suggest to use the following here instead:
> 
> 		u32 *propdata __free(kfree) = kcalloc(proplen, sizeof(*propdata), GFP_KERNEL);

Ok, so this does not work, because we jump over this causing
propdata to not get initialized, but the __free(kfree) still
triggers.

Lesson learned: do not use modern C style mixing of declarations
and statements together with __free().

Attached are updated fixup patches which do not make things crash.

Regards,

Hans




> 
>> +	if (!propdata) {
>> +		power_supply_put_battery_info(psy, info);
>> +		err = -ENOMEM;
>> +		goto out_ret_pointer;
> 
> This should be "goto out_put_node" since this is an error path.
> 
>> +	}
>> +
>> +	err = fwnode_property_read_u32_array(fwnode, "resistance-temp-table",
>> +					     propdata, proplen);
>> +	if (err < 0) {
>> +		kfree(propdata);
> 
> with the suggested "u32 *propdata __free(kfree)" this can and must be dropped.
> 
>> +		power_supply_put_battery_info(psy, info);
>> +		goto out_put_node;
>> +	}
>> +
>> +	info->resist_table_size = proplen / 2;
>>  	info->resist_table = resist_table = devm_kcalloc(&psy->dev,
>>  							 info->resist_table_size,
>>  							 sizeof(*resist_table),
>>  							 GFP_KERNEL);
>>  	if (!info->resist_table) {
>> +		kfree(propdata);
> 
> with the suggested "u32 *propdata __free(kfree)" this can and must be dropped.
> 
>>  		power_supply_put_battery_info(psy, info);
>>  		err = -ENOMEM;
>>  		goto out_put_node;
>>  	}
>>  
>>  	for (index = 0; index < info->resist_table_size; index++) {
>> -		resist_table[index].temp = be32_to_cpu(*list++);
>> -		resist_table[index].resistance = be32_to_cpu(*list++);
>> +		resist_table[index].temp = propdata[index*2];
>> +		resist_table[index].resistance = propdata[index*2+1];
>>  	}
>>  
>> +	kfree(propdata);
> 
> with the suggested "u32 *propdata __free(kfree)" this can and must be dropped.
> 
>> +
>>  out_ret_pointer:
>>  	/* Finally return the whole thing */
>>  	*info_out = info;
>>  
>>  out_put_node:
>>  	fwnode_handle_put(fwnode);
>> -	of_node_put(battery_np);
>>  	return err;
>>  }
>>  EXPORT_SYMBOL_GPL(power_supply_get_battery_info);
>>
> 
> Regards,
> 
> Hans
> 
> 

[-- Attachment #2: 0004-fixup-power-supply-core-battery-info-fully-switch-to.patch --]
[-- Type: text/x-patch, Size: 2103 bytes --]

From 68d2414d201083904dad83af4ebb2c35fc850b79 Mon Sep 17 00:00:00 2001
From: Hans de Goede <hansg@kernel.org>
Date: Sun, 8 Jun 2025 21:18:11 +0200
Subject: [PATCH 04/10] fixup! power: supply: core: battery-info: fully switch
 to fwnode

After ("power: supply: core: battery-info: fully switch to fwnode")
power_supply_get_battery_info() will always fail for battery fwnodes
which do not define a "resistance-temp-table".

Fix this by cleanly exiting on both 0 and EINVAL returns from
fwnode_property_count_u32(fwnode, "resistance-temp-table") which
indicates that the property is empty or not there.

While at it also fix:

1. The weird -ENOMEM return for other errors. For other errors
propagate the existing error or -EINVAL for an odd proplen.
2. Wrongly using "goto out_ret_pointer" on errors, out_ret_pointer
should only be used on success, error paths should use
out_put_node;

Signed-off-by: Hans de Goede <hansg@kernel.org>
---
 drivers/power/supply/power_supply_core.c | 13 +++++++------
 1 file changed, 7 insertions(+), 6 deletions(-)

diff --git a/drivers/power/supply/power_supply_core.c b/drivers/power/supply/power_supply_core.c
index a8d1fe66e248..9bbc3be2e483 100644
--- a/drivers/power/supply/power_supply_core.c
+++ b/drivers/power/supply/power_supply_core.c
@@ -795,19 +795,20 @@ int power_supply_get_battery_info(struct power_supply *psy,
 	}
 
 	proplen = fwnode_property_count_u32(fwnode, "resistance-temp-table");
-	if (proplen < 0 || proplen % 2 != 0) {
+	if (proplen == 0 || proplen == -EINVAL) {
+		err = 0;
+		goto out_ret_pointer;
+	} else if (proplen < 0 || proplen % 2 != 0) {
 		power_supply_put_battery_info(psy, info);
-		err = -ENOMEM;
-		goto out_ret_pointer;
-	} else if (proplen == 0) {
-		goto out_ret_pointer;
+		err = (proplen < 0) ? proplen : -EINVAL;
+		goto out_put_node;
 	}
 
 	propdata = kcalloc(proplen, sizeof(*propdata), GFP_KERNEL);
 	if (!propdata) {
 		power_supply_put_battery_info(psy, info);
 		err = -ENOMEM;
-		goto out_ret_pointer;
+		goto out_put_node;
 	}
 
 	err = fwnode_property_read_u32_array(fwnode, "resistance-temp-table",
-- 
2.49.0


[-- Attachment #3: 0005-fixup-power-supply-core-battery-info-fully-switch-to.patch --]
[-- Type: text/x-patch, Size: 3277 bytes --]

From 1e2d54b1c946075590d011da17d96796ede58894 Mon Sep 17 00:00:00 2001
From: Hans de Goede <hansg@kernel.org>
Date: Sun, 8 Jun 2025 21:37:59 +0200
Subject: [PATCH 05/10] fixup! power: supply: core: battery-info: fully switch
 to fwnode

propname must NOT be manually free-ed here since it is marked
__free(kfree), freeing it will cause a double-free bug.

Also move propdata to be a __free(kfree) value for consistency
and also for cleaner code.

Signed-off-by: Hans de Goede <hansg@kernel.org>
---
 drivers/power/supply/power_supply_core.c | 15 +++------------
 1 file changed, 3 insertions(+), 12 deletions(-)

diff --git a/drivers/power/supply/power_supply_core.c b/drivers/power/supply/power_supply_core.c
index 9bbc3be2e483..f2c79f15838d 100644
--- a/drivers/power/supply/power_supply_core.c
+++ b/drivers/power/supply/power_supply_core.c
@@ -588,7 +588,7 @@ int power_supply_get_battery_info(struct power_supply *psy,
 	struct fwnode_handle *srcnode, *fwnode;
 	const char *value;
 	int err, len, index, proplen;
-	u32 *propdata;
+	u32 *propdata __free(kfree) = NULL;
 	u32 min_max[2];
 
 	srcnode = dev_fwnode(&psy->dev);
@@ -758,9 +758,9 @@ int power_supply_get_battery_info(struct power_supply *psy,
 			err = -EINVAL;
 			goto out_put_node;
 		}
-		propdata = kcalloc(proplen, sizeof(*propdata), GFP_KERNEL);
+
+		u32 *propdata __free(kfree) = kcalloc(proplen, sizeof(*propdata), GFP_KERNEL);
 		if (!propdata) {
-			kfree(propname);
 			power_supply_put_battery_info(psy, info);
 			err = -EINVAL;
 			goto out_put_node;
@@ -768,8 +768,6 @@ int power_supply_get_battery_info(struct power_supply *psy,
 		err = fwnode_property_read_u32_array(fwnode, propname, propdata, proplen);
 		if (err < 0) {
 			dev_err(&psy->dev, "failed to get %s\n", propname);
-			kfree(propname);
-			kfree(propdata);
 			power_supply_put_battery_info(psy, info);
 			goto out_put_node;
 		}
@@ -780,7 +778,6 @@ int power_supply_get_battery_info(struct power_supply *psy,
 		info->ocv_table[index] = table =
 			devm_kcalloc(&psy->dev, tab_len, sizeof(*table), GFP_KERNEL);
 		if (!info->ocv_table[index]) {
-			kfree(propdata);
 			power_supply_put_battery_info(psy, info);
 			err = -ENOMEM;
 			goto out_put_node;
@@ -790,8 +787,6 @@ int power_supply_get_battery_info(struct power_supply *psy,
 			table[i].ocv = propdata[i*2];
 			table[i].capacity = propdata[i*2+1];
 		}
-
-		kfree(propdata);
 	}
 
 	proplen = fwnode_property_count_u32(fwnode, "resistance-temp-table");
@@ -814,7 +809,6 @@ int power_supply_get_battery_info(struct power_supply *psy,
 	err = fwnode_property_read_u32_array(fwnode, "resistance-temp-table",
 					     propdata, proplen);
 	if (err < 0) {
-		kfree(propdata);
 		power_supply_put_battery_info(psy, info);
 		goto out_put_node;
 	}
@@ -825,7 +819,6 @@ int power_supply_get_battery_info(struct power_supply *psy,
 							 sizeof(*resist_table),
 							 GFP_KERNEL);
 	if (!info->resist_table) {
-		kfree(propdata);
 		power_supply_put_battery_info(psy, info);
 		err = -ENOMEM;
 		goto out_put_node;
@@ -836,8 +829,6 @@ int power_supply_get_battery_info(struct power_supply *psy,
 		resist_table[index].resistance = propdata[index*2+1];
 	}
 
-	kfree(propdata);
-
 out_ret_pointer:
 	/* Finally return the whole thing */
 	*info_out = info;
-- 
2.49.0


  reply	other threads:[~2025-06-08 20:16 UTC|newest]

Thread overview: 12+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2025-04-29 22:54 [PATCH v2 0/5] power: supply: core: convert to fwnode Sebastian Reichel
2025-04-29 22:54 ` [PATCH v2 1/5] regulator: act8865-regulator: switch psy_cfg from of_node " Sebastian Reichel
2025-06-08 19:25   ` Hans de Goede
2025-04-29 22:54 ` [PATCH v2 2/5] power: supply: core: remove of_node from power_supply_config Sebastian Reichel
2025-06-08 19:26   ` Hans de Goede
2025-04-29 22:54 ` [PATCH v2 3/5] power: supply: core: battery-info: fully switch to fwnode Sebastian Reichel
2025-06-08 19:57   ` Hans de Goede
2025-06-08 20:16     ` Hans de Goede [this message]
2025-04-29 22:54 ` [PATCH v2 4/5] power: supply: core: convert to fwnnode Sebastian Reichel
2025-06-08 20:01   ` Hans de Goede
2025-04-29 22:54 ` [PATCH v2 5/5] power: supply: core: rename power_supply_get_by_phandle to power_supply_get_by_reference Sebastian Reichel
2025-06-08 20:25   ` Hans de Goede

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=c76fe3f8-8884-4769-af69-a9a527d2a202@kernel.org \
    --to=hansg@kernel.org \
    --cc=angelogioacchino.delregno@collabora.com \
    --cc=broonie@kernel.org \
    --cc=jernej.skrabec@gmail.com \
    --cc=kishon@kernel.org \
    --cc=krzk@kernel.org \
    --cc=lgirdwood@gmail.com \
    --cc=linus.walleij@linaro.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-pm@vger.kernel.org \
    --cc=mazziesaccount@gmail.com \
    --cc=pali@kernel.org \
    --cc=samuel@sholland.org \
    --cc=sebastian.reichel@collabora.com \
    --cc=sre@kernel.org \
    --cc=vkoul@kernel.org \
    --cc=wens@csie.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).