dmaengine: acpi: Free resource list at appropriate time
PATCH 1/1 dmaengine: acpi: Free resource list at appropriate time
In one case we don't free resources when formally should, and in the other we do unneeded "double free" (emptying an empty list).
Both are not critical issues at all, they just make code robust against any possible future changes in the flow.
--- drivers/dma/acpi-dma.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-)
drivers/dma/acpi-dma.c +3 -2
@@ -55,7 +55,7 @@ static int acpi_dma_parse_resource_group(const struct acpi_csrt_group *grp,INIT_LIST_HEAD (&resource_list); ret =acpi_dev_get_resources (adev, &resource_list,NULL ,NULL ); -if (ret <=0 ) +if (ret <0 )return 0 ;list_for_each_entry (rentry, &resource_list, node) {
@@ -370,10 +370,11 @@ struct dma_chan *acpi_dma_request_slave_chan_by_index(struct device *dev,INIT_LIST_HEAD (&resource_list); ret =acpi_dev_get_resources (adev, &resource_list, acpi_dma_parse_fixed_dma, &pdata); -acpi_dev_free_resource_list (&resource_list);if (ret <0 )return ERR_PTR (ret); +acpi_dev_free_resource_list (&resource_list); +if (dma_spec->slave_id <0 || dma_spec->chan_id <0 )return ERR_PTR (-ENODEV );
Signature · 1 line
2.50.1
Quoted 6 lines · Andy Shevchenko
In one case we don't free resources when formally should, and in the other we do unneeded "double free" (emptying an empty list).
Both are not critical issues at all, they just make code robust against any possible future changes in the flow.
what are you talking about, can use straight forward words.
You just move acpi_dev_free_resource_list() after check return value of acpi_dev_get_resources().
Quoted 11 lines
Signed-off-by Andy Shevchenko <[email protected]>--- drivers/dma/acpi-dma.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-)drivers/dma/acpi-dma.c +0
-0
drivers/dma/acpi-dma.c static int acpi_dma_parse_resource_group(const struct acpi_csrt_group *grp, · line 59
Quoted 6 lines
drivers/dma/acpi-dma.c +1
-1@@ -55,7 +55,7 @@ static int acpi_dma_parse_resource_group(const struct acpi_csrt_group *grp,INIT_LIST_HEAD (&resource_list); ret =acpi_dev_get_resources (adev, &resource_list,NULL ,NULL ); -if (ret <=0 ) +if (ret <0 )return 0 ;
Dose this change related with this patch?
Frank
Quoted 18 lines
drivers/dma/acpi-dma.c +2
-2list_for_each_entry (rentry, &resource_list, node) {@@ -370,10 +370,11 @@ struct dma_chan *acpi_dma_request_slave_chan_by_index(struct device *dev,INIT_LIST_HEAD (&resource_list); ret =acpi_dev_get_resources (adev, &resource_list, acpi_dma_parse_fixed_dma, &pdata); -acpi_dev_free_resource_list (&resource_list);if (ret <0 )return ERR_PTR (ret); +acpi_dev_free_resource_list (&resource_list); +if (dma_spec->slave_id <0 || dma_spec->chan_id <0 )return ERR_PTR (-ENODEV ); --2.50.1
Quoted 5 lines · Frank Li
Quoted 4 lines · Andy Shevchenko
In one case we don't free resources when formally should, and in the other we do unneeded "double free" (emptying an empty list).
Quoted 7 lines
Quoted 2 lines
Both are not critical issues at all, they just make code robust against any possible future changes in the flow.
what are you talking about, can use straight forward words.
You just move
acpi_dev_free_resource_list()after check return value ofacpi_dev_get_resources().
Not really. There are _two_ cases. And while they are slightly different, the solution is to make them behave in the same ways.
...
Quoted 7 lines
drivers/dma/acpi-dma.c static int acpi_dma_parse_resource_group(const struct acpi_csrt_group *grp, · line 59
Quoted 5 lines
drivers/dma/acpi-dma.c +1
-1INIT_LIST_HEAD (&resource_list); ret =acpi_dev_get_resources (adev, &resource_list,NULL ,NULL ); -if (ret <=0 ) +if (ret <0 )return 0 ;Dose this change related with this patch?
Yes. It's mentioned in the very first part of the first sentence in the commit message.
Quoted 18 lines
Quoted 18 lines
drivers/dma/acpi-dma.c +2
-2list_for_each_entry (rentry, &resource_list, node) {@@ -370,10 +370,11 @@ struct dma_chan *acpi_dma_request_slave_chan_by_index(struct device *dev,INIT_LIST_HEAD (&resource_list); ret =acpi_dev_get_resources (adev, &resource_list, acpi_dma_parse_fixed_dma, &pdata); -acpi_dev_free_resource_list (&resource_list);if (ret <0 )return ERR_PTR (ret); +acpi_dev_free_resource_list (&resource_list); +if (dma_spec->slave_id <0 || dma_spec->chan_id <0 )return ERR_PTR (-ENODEV ); --2.50.1
Signature · 2 lines
With Best Regards, Andy Shevchenko
Quoted 6 lines · Andy Shevchenko
In one case we don't free resources when formally should, and in the other we do unneeded "double free" (emptying an empty list).
Both are not critical issues at all, they just make code robust against any possible future changes in the flow.
Can you rephrase commit message
In acpi_dma_parse_resource_group(), change the return condition from "ret <= 0" to "ret < 0" so that when acpi_dev_get_resources() returns 0 (no resources), acpi_dev_free_resource_list() is still called to free the empty list instead of returning early.
In acpi_dma_request_slave_chan_by_index(), move acpi_dev_free_resource_list() to after the error check so that on the error return path the resource list is properly freed.
Frank
Quoted 35 lines
Signed-off-by Andy Shevchenko <[email protected]>--- drivers/dma/acpi-dma.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-)drivers/dma/acpi-dma.c +3
-3@@ -55,7 +55,7 @@ static int acpi_dma_parse_resource_group(const struct acpi_csrt_group *grp,INIT_LIST_HEAD (&resource_list); ret =acpi_dev_get_resources (adev, &resource_list,NULL ,NULL ); -if (ret <=0 ) +if (ret <0 )return 0 ;list_for_each_entry (rentry, &resource_list, node) {@@ -370,10 +370,11 @@ struct dma_chan *acpi_dma_request_slave_chan_by_index(struct device *dev,INIT_LIST_HEAD (&resource_list); ret =acpi_dev_get_resources (adev, &resource_list, acpi_dma_parse_fixed_dma, &pdata); -acpi_dev_free_resource_list (&resource_list);if (ret <0 )return ERR_PTR (ret); +acpi_dev_free_resource_list (&resource_list); +if (dma_spec->slave_id <0 || dma_spec->chan_id <0 )return ERR_PTR (-ENODEV ); --2.50.1