[PATCH] ACPI / DMAR: Avoid passing NULL to acpi_put_table()

[PATCH] ACPI / DMAR: Avoid passing NULL to acpi_put_table()

[Rafael J. Wysocki]

From: Rafael J. Wysocki <rafael.j.wysocki@intel.com>

Linus reported that commit 174cc7187e6f "ACPICA: Tables: Back port
acpi_get_table_with_size() and early_acpi_os_unmap_memory() from
Linux kernel" added a new warning on his desktop system:

 ACPI Warning: Table ffffffff9fe6c0a0, Validation count is zero before decrement

which turns out to come from the acpi_put_table() in
detect_intel_iommu().

This happens if the DMAR table is not present in which case NULL is
passed to acpi_put_table() which doesn't check against that and
attempts to handle it regardless.

For this reason, check the pointer passed to acpi_put_table()
before invoking it.

Reported-by: Linus Torvalds <torvalds@linux-foundation.org> 
Fixes: 6b11d1d67713 ("ACPI / osl: Remove acpi_get_table_with_size()/early_acpi_os_unmap_memory() users")
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
---
 drivers/iommu/dmar.c |    6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

Index: linux-pm/drivers/iommu/dmar.c
===================================================================
--- linux-pm.orig/drivers/iommu/dmar.c
+++ linux-pm/drivers/iommu/dmar.c
@@ -903,8 +903,10 @@ int __init detect_intel_iommu(void)
 		x86_init.iommu.iommu_init = intel_iommu_init;
 #endif
 
-	acpi_put_table(dmar_tbl);
-	dmar_tbl = NULL;
+	if (dmar_tbl) {
+		acpi_put_table(dmar_tbl);
+		dmar_tbl = NULL;
+	}
 	up_write(&dmar_global_lock);
 
 	return ret ? 1 : -ENODEV;

Re: [PATCH] ACPI / DMAR: Avoid passing NULL to acpi_put_table()

[Paul E. McKenney]

On Thu, Jan 05, 2017 at 01:24:14AM +0100, Rafael J. Wysocki wrote:
> From: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
> 
> Linus reported that commit 174cc7187e6f "ACPICA: Tables: Back port
> acpi_get_table_with_size() and early_acpi_os_unmap_memory() from
> Linux kernel" added a new warning on his desktop system:
> 
>  ACPI Warning: Table ffffffff9fe6c0a0, Validation count is zero before decrement
> 
> which turns out to come from the acpi_put_table() in
> detect_intel_iommu().
> 
> This happens if the DMAR table is not present in which case NULL is
> passed to acpi_put_table() which doesn't check against that and
> attempts to handle it regardless.
> 
> For this reason, check the pointer passed to acpi_put_table()
> before invoking it.
> 
> Reported-by: Linus Torvalds <torvalds@linux-foundation.org> 
> Fixes: 6b11d1d67713 ("ACPI / osl: Remove acpi_get_table_with_size()/early_acpi_os_unmap_memory() users")
> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>

Much better!

Tested-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>

> ---
>  drivers/iommu/dmar.c |    6 ++++--
>  1 file changed, 4 insertions(+), 2 deletions(-)
> 
> Index: linux-pm/drivers/iommu/dmar.c
> ===================================================================
> --- linux-pm.orig/drivers/iommu/dmar.c
> +++ linux-pm/drivers/iommu/dmar.c
> @@ -903,8 +903,10 @@ int __init detect_intel_iommu(void)
>  		x86_init.iommu.iommu_init = intel_iommu_init;
>  #endif
> 
> -	acpi_put_table(dmar_tbl);
> -	dmar_tbl = NULL;
> +	if (dmar_tbl) {
> +		acpi_put_table(dmar_tbl);
> +		dmar_tbl = NULL;
> +	}
>  	up_write(&dmar_global_lock);
> 
>  	return ret ? 1 : -ENODEV;
> 

Re: [PATCH] ACPI / DMAR: Avoid passing NULL to acpi_put_table()

[Andy Shevchenko]

On Thu, Jan 5, 2017 at 2:24 AM, Rafael J. Wysocki <rjw-LthD3rsA81gm4RdzfppkhA@public.gmane.org> wrote:
> From: Rafael J. Wysocki <rafael.j.wysocki-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
>
> Linus reported that commit 174cc7187e6f "ACPICA: Tables: Back port
> acpi_get_table_with_size() and early_acpi_os_unmap_memory() from
> Linux kernel" added a new warning on his desktop system:
>
>  ACPI Warning: Table ffffffff9fe6c0a0, Validation count is zero before decrement
>
> which turns out to come from the acpi_put_table() in
> detect_intel_iommu().
>
> This happens if the DMAR table is not present in which case NULL is
> passed to acpi_put_table() which doesn't check against that and
> attempts to handle it regardless.
>
> For this reason, check the pointer passed to acpi_put_table()
> before invoking it.
>

Reviewed-by: Andy Shevchenko <andy.shevchenko-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>

> Reported-by: Linus Torvalds <torvalds-de/tnXTf+JLsfHDXvbKv3WD2FQJk+8+b@public.gmane.org>
> Fixes: 6b11d1d67713 ("ACPI / osl: Remove acpi_get_table_with_size()/early_acpi_os_unmap_memory() users")
> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
> ---
>  drivers/iommu/dmar.c |    6 ++++--
>  1 file changed, 4 insertions(+), 2 deletions(-)
>
> Index: linux-pm/drivers/iommu/dmar.c
> ===================================================================
> --- linux-pm.orig/drivers/iommu/dmar.c
> +++ linux-pm/drivers/iommu/dmar.c
> @@ -903,8 +903,10 @@ int __init detect_intel_iommu(void)
>                 x86_init.iommu.iommu_init = intel_iommu_init;
>  #endif
>
> -       acpi_put_table(dmar_tbl);
> -       dmar_tbl = NULL;
> +       if (dmar_tbl) {
> +               acpi_put_table(dmar_tbl);
> +               dmar_tbl = NULL;
> +       }
>         up_write(&dmar_global_lock);
>
>         return ret ? 1 : -ENODEV;
>



-- 
With Best Regards,
Andy Shevchenko

RE: [PATCH] ACPI / DMAR: Avoid passing NULL to acpi_put_table()

[Zheng, Lv]

Hi, Rafael

> From: linux-acpi-owner-u79uwXL29TY76Z2rM5mHXA@public.gmane.org [mailto:linux-acpi-owner-u79uwXL29TY76Z2rM5mHXA@public.gmane.org] On Behalf Of Rafael J.
> Wysocki
> Subject: [PATCH] ACPI / DMAR: Avoid passing NULL to acpi_put_table()
> 
> From: Rafael J. Wysocki <rafael.j.wysocki-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
> 
> Linus reported that commit 174cc7187e6f "ACPICA: Tables: Back port
> acpi_get_table_with_size() and early_acpi_os_unmap_memory() from
> Linux kernel" added a new warning on his desktop system:
> 
>  ACPI Warning: Table ffffffff9fe6c0a0, Validation count is zero before decrement
> 
> which turns out to come from the acpi_put_table() in
> detect_intel_iommu().
> 
> This happens if the DMAR table is not present in which case NULL is
> passed to acpi_put_table() which doesn't check against that and
> attempts to handle it regardless.
> 
> For this reason, check the pointer passed to acpi_put_table()
> before invoking it.
> 
> Reported-by: Linus Torvalds <torvalds-de/tnXTf+JLsfHDXvbKv3WD2FQJk+8+b@public.gmane.org>
> Fixes: 6b11d1d67713 ("ACPI / osl: Remove acpi_get_table_with_size()/early_acpi_os_unmap_memory()
> users")
> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>

Thanks for looking into this issue.
The patch fixes the caller side and thus looks very good from my opinion.
Acked-by: Lv Zheng <lv.zheng-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>

Best regards
Lv

> ---
>  drivers/iommu/dmar.c |    6 ++++--
>  1 file changed, 4 insertions(+), 2 deletions(-)
> 
> Index: linux-pm/drivers/iommu/dmar.c
> ===================================================================
> --- linux-pm.orig/drivers/iommu/dmar.c
> +++ linux-pm/drivers/iommu/dmar.c
> @@ -903,8 +903,10 @@ int __init detect_intel_iommu(void)
>  		x86_init.iommu.iommu_init = intel_iommu_init;
>  #endif
> 
> -	acpi_put_table(dmar_tbl);
> -	dmar_tbl = NULL;
> +	if (dmar_tbl) {
> +		acpi_put_table(dmar_tbl);
> +		dmar_tbl = NULL;
> +	}
>  	up_write(&dmar_global_lock);
> 
>  	return ret ? 1 : -ENODEV;
> 
> --
> To unsubscribe from this list: send the line "unsubscribe linux-acpi" in
> the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html