* [PATCH v2] PCI/ACPI: Fix double free bug in pci_acpi_scan_root() function
@ 2025-06-17 2:37 Zhe Qiao
2025-06-17 14:40 ` Bjorn Helgaas
2025-06-18 16:02 ` Dan Carpenter
0 siblings, 2 replies; 4+ messages in thread
From: Zhe Qiao @ 2025-06-17 2:37 UTC (permalink / raw)
To: rafael
Cc: bhelgaas, lenb, kwilczynski, sashal, linux-kernel, linux-pci,
linux-acpi, qiaozhe
The patch "PCI/ACPI: Fix allocated memory release on error in
pci_acpi_scan_root()" introduces a dual release issue. When
acpi_pci_root_creat() fails, the pci_cpi_can_root() function
will release 'ri ->cfg' and 'root_ops' in the error handling
path.However, acpi_pci_root_creat() will also call
__acpi_pci_root_release_info(), which in turn will call the
release_info hook, causing the same block of memory to be
released again.
Fixes: 631b2af2f357 ("PCI/ACPI: Fix allocated memory release on error in pci_acpi_scan_root()")
Signed-off-by: Zhe Qiao <qiaozhe@iscas.ac.cn>
---
v1 -> v2:
- Restore all changes from the first version.
- Remove unnecessary release info hooks.
- Add a NULL check before calling info->ops->release_info().
- Delete the currently unused pci_api_geneic_delease_info () function.
---
drivers/acpi/pci_root.c | 3 ++-
drivers/pci/pci-acpi.c | 12 ------------
2 files changed, 2 insertions(+), 13 deletions(-)
diff --git a/drivers/acpi/pci_root.c b/drivers/acpi/pci_root.c
index 74ade4160314..83628adbc56b 100644
--- a/drivers/acpi/pci_root.c
+++ b/drivers/acpi/pci_root.c
@@ -974,7 +974,8 @@ static void __acpi_pci_root_release_info(struct acpi_pci_root_info *info)
resource_list_destroy_entry(entry);
}
- info->ops->release_info(info);
+ if (info->ops && info->ops->release_info)
+ info->ops->release_info(info);
}
static void acpi_pci_root_release_info(struct pci_host_bridge *bridge)
diff --git a/drivers/pci/pci-acpi.c b/drivers/pci/pci-acpi.c
index b78e0e417324..6e85816ee1c3 100644
--- a/drivers/pci/pci-acpi.c
+++ b/drivers/pci/pci-acpi.c
@@ -1652,17 +1652,6 @@ pci_acpi_setup_ecam_mapping(struct acpi_pci_root *root)
return cfg;
}
-/* release_info: free resources allocated by init_info */
-static void pci_acpi_generic_release_info(struct acpi_pci_root_info *ci)
-{
- struct acpi_pci_generic_root_info *ri;
-
- ri = container_of(ci, struct acpi_pci_generic_root_info, common);
- pci_ecam_free(ri->cfg);
- kfree(ci->ops);
- kfree(ri);
-}
-
/* Interface called from ACPI code to setup PCI host controller */
struct pci_bus *pci_acpi_scan_root(struct acpi_pci_root *root)
{
@@ -1683,7 +1672,6 @@ struct pci_bus *pci_acpi_scan_root(struct acpi_pci_root *root)
if (!ri->cfg)
goto free_root_ops;
- root_ops->release_info = pci_acpi_generic_release_info;
root_ops->prepare_resources = pci_acpi_root_prepare_resources;
root_ops->pci_ops = (struct pci_ops *)&ri->cfg->ops->pci_ops;
bus = acpi_pci_root_create(root, root_ops, &ri->common, ri->cfg);
--
2.43.0
^ permalink raw reply related [flat|nested] 4+ messages in thread
* Re: [PATCH v2] PCI/ACPI: Fix double free bug in pci_acpi_scan_root() function
2025-06-17 2:37 [PATCH v2] PCI/ACPI: Fix double free bug in pci_acpi_scan_root() function Zhe Qiao
@ 2025-06-17 14:40 ` Bjorn Helgaas
2025-06-18 2:24 ` Zhe Qiao
2025-06-18 16:02 ` Dan Carpenter
1 sibling, 1 reply; 4+ messages in thread
From: Bjorn Helgaas @ 2025-06-17 14:40 UTC (permalink / raw)
To: Zhe Qiao
Cc: rafael, bhelgaas, lenb, kwilczynski, sashal, linux-kernel,
linux-pci, linux-acpi, Dan Carpenter
[+cc Dan]
On Tue, Jun 17, 2025 at 10:37:38AM +0800, Zhe Qiao wrote:
> The patch "PCI/ACPI: Fix allocated memory release on error in
> pci_acpi_scan_root()" introduces a dual release issue. When
> acpi_pci_root_creat() fails, the pci_cpi_can_root() function
> will release 'ri ->cfg' and 'root_ops' in the error handling
> path.However, acpi_pci_root_creat() will also call
> __acpi_pci_root_release_info(), which in turn will call the
> release_info hook, causing the same block of memory to be
> released again.
These are all nits, but would have to be fixed before applying:
- 'The patch "PCI/ACPI: Fix ..."' is not the usual way to identify a
commit. Use the same style as in the Fixes: tag below.
- Typo in "acpi_pci_root_creat" (twice)
- Typo in "pci_cpi_can_root"
- Add space after the period in "path.However, ..."
- Add "Reported-by: Dan Carpenter <dan.carpenter@linaro.org>" and
"Closes: https://lore.kernel.org/all/aEmdnuw715btq7Q5@stanley.mountain/"
and cc: Dan.
- 631b2af2f357 appeared in v6.16-rc1, so we should try to get the
fix into v6.16. A hint after the "---" would be helpful to make
sure that happens.
Wait a few days before reposting in case other folks have comments.
> Fixes: 631b2af2f357 ("PCI/ACPI: Fix allocated memory release on error in pci_acpi_scan_root()")
> Signed-off-by: Zhe Qiao <qiaozhe@iscas.ac.cn>
> ---
> v1 -> v2:
> - Restore all changes from the first version.
> - Remove unnecessary release info hooks.
> - Add a NULL check before calling info->ops->release_info().
> - Delete the currently unused pci_api_geneic_delease_info () function.
> ---
> drivers/acpi/pci_root.c | 3 ++-
> drivers/pci/pci-acpi.c | 12 ------------
> 2 files changed, 2 insertions(+), 13 deletions(-)
>
> diff --git a/drivers/acpi/pci_root.c b/drivers/acpi/pci_root.c
> index 74ade4160314..83628adbc56b 100644
> --- a/drivers/acpi/pci_root.c
> +++ b/drivers/acpi/pci_root.c
> @@ -974,7 +974,8 @@ static void __acpi_pci_root_release_info(struct acpi_pci_root_info *info)
> resource_list_destroy_entry(entry);
> }
>
> - info->ops->release_info(info);
> + if (info->ops && info->ops->release_info)
> + info->ops->release_info(info);
> }
>
> static void acpi_pci_root_release_info(struct pci_host_bridge *bridge)
> diff --git a/drivers/pci/pci-acpi.c b/drivers/pci/pci-acpi.c
> index b78e0e417324..6e85816ee1c3 100644
> --- a/drivers/pci/pci-acpi.c
> +++ b/drivers/pci/pci-acpi.c
> @@ -1652,17 +1652,6 @@ pci_acpi_setup_ecam_mapping(struct acpi_pci_root *root)
> return cfg;
> }
>
> -/* release_info: free resources allocated by init_info */
> -static void pci_acpi_generic_release_info(struct acpi_pci_root_info *ci)
> -{
> - struct acpi_pci_generic_root_info *ri;
> -
> - ri = container_of(ci, struct acpi_pci_generic_root_info, common);
> - pci_ecam_free(ri->cfg);
> - kfree(ci->ops);
> - kfree(ri);
> -}
> -
> /* Interface called from ACPI code to setup PCI host controller */
> struct pci_bus *pci_acpi_scan_root(struct acpi_pci_root *root)
> {
> @@ -1683,7 +1672,6 @@ struct pci_bus *pci_acpi_scan_root(struct acpi_pci_root *root)
> if (!ri->cfg)
> goto free_root_ops;
>
> - root_ops->release_info = pci_acpi_generic_release_info;
> root_ops->prepare_resources = pci_acpi_root_prepare_resources;
> root_ops->pci_ops = (struct pci_ops *)&ri->cfg->ops->pci_ops;
> bus = acpi_pci_root_create(root, root_ops, &ri->common, ri->cfg);
> --
> 2.43.0
>
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [PATCH v2] PCI/ACPI: Fix double free bug in pci_acpi_scan_root() function
2025-06-17 14:40 ` Bjorn Helgaas
@ 2025-06-18 2:24 ` Zhe Qiao
0 siblings, 0 replies; 4+ messages in thread
From: Zhe Qiao @ 2025-06-18 2:24 UTC (permalink / raw)
To: Bjorn Helgaas
Cc: rafael, bhelgaas, lenb, kwilczynski, sashal, linux-kernel,
linux-pci, linux-acpi, Dan Carpenter
Hi Bjorn
On Tue, Jun 17, 2025 at 10:40 PM Bjorn Helgaas <helgaas@kernel.org> wrote:
>
> [+cc Dan]
>
> On Tue, Jun 17, 2025 at 10:37:38AM +0800, Zhe Qiao wrote:
> > The patch "PCI/ACPI: Fix allocated memory release on error in
> > pci_acpi_scan_root()" introduces a dual release issue. When
> > acpi_pci_root_creat() fails, the pci_cpi_can_root() function
> > will release 'ri ->cfg' and 'root_ops' in the error handling
> > path.However, acpi_pci_root_creat() will also call
> > __acpi_pci_root_release_info(), which in turn will call the
> > release_info hook, causing the same block of memory to be
> > released again.
>
> These are all nits, but would have to be fixed before applying:
>
> - 'The patch "PCI/ACPI: Fix ..."' is not the usual way to identify a
> commit. Use the same style as in the Fixes: tag below.
>
> - Typo in "acpi_pci_root_creat" (twice)
>
> - Typo in "pci_cpi_can_root"
>
> - Add space after the period in "path.However, ..."
>
> - Add "Reported-by: Dan Carpenter <dan.carpenter@linaro.org>" and
> "Closes: https://lore.kernel.org/all/aEmdnuw715btq7Q5@stanley.mountain/"
> and cc: Dan.
>
> - 631b2af2f357 appeared in v6.16-rc1, so we should try to get the
> fix into v6.16. A hint after the "---" would be helpful to make
> sure that happens.
>
> Wait a few days before reposting in case other folks have comments.
>
Thank you for your detailed review and pointing out these issues.
I will wait a few days before resending.
Thanks again!
> > Fixes: 631b2af2f357 ("PCI/ACPI: Fix allocated memory release on error in pci_acpi_scan_root()")
> > Signed-off-by: Zhe Qiao <qiaozhe@iscas.ac.cn>
> > ---
> > v1 -> v2:
> > - Restore all changes from the first version.
> > - Remove unnecessary release info hooks.
> > - Add a NULL check before calling info->ops->release_info().
> > - Delete the currently unused pci_api_geneic_delease_info () function.
> > ---
> > drivers/acpi/pci_root.c | 3 ++-
> > drivers/pci/pci-acpi.c | 12 ------------
> > 2 files changed, 2 insertions(+), 13 deletions(-)
> >
> > diff --git a/drivers/acpi/pci_root.c b/drivers/acpi/pci_root.c
> > index 74ade4160314..83628adbc56b 100644
> > --- a/drivers/acpi/pci_root.c
> > +++ b/drivers/acpi/pci_root.c
> > @@ -974,7 +974,8 @@ static void __acpi_pci_root_release_info(struct acpi_pci_root_info *info)
> > resource_list_destroy_entry(entry);
> > }
> >
> > - info->ops->release_info(info);
> > + if (info->ops && info->ops->release_info)
> > + info->ops->release_info(info);
> > }
> >
> > static void acpi_pci_root_release_info(struct pci_host_bridge *bridge)
> > diff --git a/drivers/pci/pci-acpi.c b/drivers/pci/pci-acpi.c
> > index b78e0e417324..6e85816ee1c3 100644
> > --- a/drivers/pci/pci-acpi.c
> > +++ b/drivers/pci/pci-acpi.c
> > @@ -1652,17 +1652,6 @@ pci_acpi_setup_ecam_mapping(struct acpi_pci_root *root)
> > return cfg;
> > }
> >
> > -/* release_info: free resources allocated by init_info */
> > -static void pci_acpi_generic_release_info(struct acpi_pci_root_info *ci)
> > -{
> > - struct acpi_pci_generic_root_info *ri;
> > -
> > - ri = container_of(ci, struct acpi_pci_generic_root_info, common);
> > - pci_ecam_free(ri->cfg);
> > - kfree(ci->ops);
> > - kfree(ri);
> > -}
> > -
> > /* Interface called from ACPI code to setup PCI host controller */
> > struct pci_bus *pci_acpi_scan_root(struct acpi_pci_root *root)
> > {
> > @@ -1683,7 +1672,6 @@ struct pci_bus *pci_acpi_scan_root(struct acpi_pci_root *root)
> > if (!ri->cfg)
> > goto free_root_ops;
> >
> > - root_ops->release_info = pci_acpi_generic_release_info;
> > root_ops->prepare_resources = pci_acpi_root_prepare_resources;
> > root_ops->pci_ops = (struct pci_ops *)&ri->cfg->ops->pci_ops;
> > bus = acpi_pci_root_create(root, root_ops, &ri->common, ri->cfg);
> > --
> > 2.43.0
> >
regards,
Zhe
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [PATCH v2] PCI/ACPI: Fix double free bug in pci_acpi_scan_root() function
2025-06-17 2:37 [PATCH v2] PCI/ACPI: Fix double free bug in pci_acpi_scan_root() function Zhe Qiao
2025-06-17 14:40 ` Bjorn Helgaas
@ 2025-06-18 16:02 ` Dan Carpenter
1 sibling, 0 replies; 4+ messages in thread
From: Dan Carpenter @ 2025-06-18 16:02 UTC (permalink / raw)
To: Zhe Qiao
Cc: rafael, bhelgaas, lenb, kwilczynski, sashal, linux-kernel,
linux-pci, linux-acpi
Wait, what? We can't just delete pci_acpi_generic_release_info().
What's going to free all those things on the success path? I
don't understand.
I had suggested you do something like this:
diff --git a/drivers/acpi/pci_root.c b/drivers/acpi/pci_root.c
index 74ade4160314..ff5799b696d6 100644
--- a/drivers/acpi/pci_root.c
+++ b/drivers/acpi/pci_root.c
@@ -1012,20 +1012,20 @@ struct pci_bus *acpi_pci_root_create(struct acpi_pci_root *root,
root->segment, busnum);
if (ops->init_info && ops->init_info(info))
- goto out_release_info;
+ return NULL;
if (ops->prepare_resources)
ret = ops->prepare_resources(info);
else
ret = acpi_pci_probe_root_resources(info);
if (ret < 0)
- goto out_release_info;
+ goto cleanup_init_info;
pci_acpi_root_add_resources(info);
pci_add_resource(&info->resources, &root->secondary);
bus = pci_create_root_bus(NULL, busnum, ops->pci_ops,
sysdata, &info->resources);
if (!bus)
- goto out_release_info;
+ goto cleanup_acpi_pci_probe_root_resources;
host_bridge = to_pci_host_bridge(bus->bridge);
if (!(root->osc_control_set & OSC_PCI_EXPRESS_NATIVE_HP_CONTROL))
@@ -1053,8 +1053,10 @@ struct pci_bus *acpi_pci_root_create(struct acpi_pci_root *root,
dev_printk(KERN_DEBUG, &bus->dev, "on NUMA node %d\n", node);
return bus;
+cleanup_acpi_pci_probe_root_resources:
+ free_something_with_probe();
out_release_info:
- __acpi_pci_root_release_info(info);
+ free_something_info();
return NULL;
}
But you'd need to replace the dummy code with actual code and
change the callers etc. And I haven't looked at the actual
code to verify it's a good idea.
regards,
dan carpenter
^ permalink raw reply related [flat|nested] 4+ messages in thread
end of thread, other threads:[~2025-06-18 16:02 UTC | newest]
Thread overview: 4+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2025-06-17 2:37 [PATCH v2] PCI/ACPI: Fix double free bug in pci_acpi_scan_root() function Zhe Qiao
2025-06-17 14:40 ` Bjorn Helgaas
2025-06-18 2:24 ` Zhe Qiao
2025-06-18 16:02 ` Dan Carpenter
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).