2023-02-24 02:38:26

by Weichen Chen

Subject: [PATCH v1 1/1] pstore/ram: Fix crash when setting number of cpus to an odd number

When the number of cpu cores is adjusted to 7 or other odd numbers,
the zone size will become an odd number.
The address of the zone will become:
addr of zone0 = BASE
addr of zone1 = BASE + zone_size
addr of zone2 = BASE + zone_size*2
...
The address of zone1/3/5/7 will be mapped to non-alignment va.
Eventually crashes will occur when accessing these va.

So, use ALIGN_DOWN() to make sure the zone size is even
to avoid this bug.

Signed-off-by: Weichen Chen <[email protected]>
---
fs/pstore/ram.c | 1 +
1 file changed, 1 insertion(+)

diff --git a/fs/pstore/ram.c b/fs/pstore/ram.c
index ade66dbe5f39..fc57ac97e506 100644
--- a/fs/pstore/ram.c
+++ b/fs/pstore/ram.c
@@ -528,6 +528,7 @@ static int ramoops_init_przs(const char *name,
}

zone_sz = mem_sz / *cnt;
+ zone_sz = ALIGN_DOWN(zone_sz, 2);
if (!zone_sz) {
dev_err(dev, "%s zone size == 0\n", name);
goto fail;
--
2.18.0

2023-02-24 04:01:40

by Guilherme G. Piccoli

Subject: Re: [PATCH v1 1/1] pstore/ram: Fix crash when setting number of cpus to an odd number

On 23/02/2023 23:36, Weichen Chen wrote:
> When the number of cpu cores is adjusted to 7 or other odd numbers,
> the zone size will become an odd number.
> The address of the zone will become:
> addr of zone0 = BASE
> addr of zone1 = BASE + zone_size
> addr of zone2 = BASE + zone_size*2
> ...
> The address of zone1/3/5/7 will be mapped to non-alignment va.
> Eventually crashes will occur when accessing these va.
>
> So, use ALIGN_DOWN() to make sure the zone size is even
> to avoid this bug.
>
> Signed-off-by: Weichen Chen <[email protected]>
> ---
> fs/pstore/ram.c | 1 +
> 1 file changed, 1 insertion(+)
>
> diff --git a/fs/pstore/ram.c b/fs/pstore/ram.c
> index ade66dbe5f39..fc57ac97e506 100644
> --- a/fs/pstore/ram.c
> +++ b/fs/pstore/ram.c
> @@ -528,6 +528,7 @@ static int ramoops_init_przs(const char *name,
> }
>
> zone_sz = mem_sz / *cnt;
> + zone_sz = ALIGN_DOWN(zone_sz, 2);
> if (!zone_sz) {
> dev_err(dev, "%s zone size == 0\n", name);
> goto fail;

Thanks for resending! Feel free to add my:

Tested-by: Guilherme G. Piccoli <[email protected]>

Cheers,


Guilherme

2023-02-24 07:05:13

by Miko Larsson

Subject: Re: [PATCH v1 1/1] pstore/ram: Fix crash when setting number of cpus to an odd number

On Fri, 2023-02-24 at 10:36 +0800, Weichen Chen wrote:
> When the number of cpu cores is adjusted to 7 or other odd numbers,
> the zone size will become an odd number.
> The address of the zone will become:
>     addr of zone0 = BASE
>     addr of zone1 = BASE + zone_size
>     addr of zone2 = BASE + zone_size*2
>     ...
> The address of zone1/3/5/7 will be mapped to non-alignment va.
> Eventually crashes will occur when accessing these va.
>
> So, use ALIGN_DOWN() to make sure the zone size is even
> to avoid this bug.
>
> Signed-off-by: Weichen Chen <[email protected]>
> ---
>  fs/pstore/ram.c | 1 +
>  1 file changed, 1 insertion(+)
>
> diff --git a/fs/pstore/ram.c b/fs/pstore/ram.c
> index ade66dbe5f39..fc57ac97e506 100644
> --- a/fs/pstore/ram.c
> +++ b/fs/pstore/ram.c
> @@ -528,6 +528,7 @@ static int ramoops_init_przs(const char *name,
>         }
>  
>         zone_sz = mem_sz / *cnt;
> +       zone_sz = ALIGN_DOWN(zone_sz, 2);
>         if (!zone_sz) {
>                 dev_err(dev, "%s zone size == 0\n", name);
>                 goto fail;

Might want to Cc this to the stable mailing list.
--
~miko

2023-02-24 10:09:39

by Matthias Brugger

Subject: Re: [PATCH v1 1/1] pstore/ram: Fix crash when setting number of cpus to an odd number

On 24/02/2023 05:00, Guilherme G. Piccoli wrote:
> On 23/02/2023 23:36, Weichen Chen wrote:
>> When the number of cpu cores is adjusted to 7 or other odd numbers,
>> the zone size will become an odd number.
>> The address of the zone will become:
>> addr of zone0 = BASE
>> addr of zone1 = BASE + zone_size
>> addr of zone2 = BASE + zone_size*2
>> ...
>> The address of zone1/3/5/7 will be mapped to non-alignment va.
>> Eventually crashes will occur when accessing these va.
>>
>> So, use ALIGN_DOWN() to make sure the zone size is even
>> to avoid this bug.
>>
>> Signed-off-by: Weichen Chen <[email protected]>
>> ---
>> fs/pstore/ram.c | 1 +
>> 1 file changed, 1 insertion(+)
>>
>> diff --git a/fs/pstore/ram.c b/fs/pstore/ram.c
>> index ade66dbe5f39..fc57ac97e506 100644
>> --- a/fs/pstore/ram.c
>> +++ b/fs/pstore/ram.c
>> @@ -528,6 +528,7 @@ static int ramoops_init_przs(const char *name,
>> }
>>
>> zone_sz = mem_sz / *cnt;
>> + zone_sz = ALIGN_DOWN(zone_sz, 2);
>> if (!zone_sz) {
>> dev_err(dev, "%s zone size == 0\n", name);
>> goto fail;
>
> Thanks for resending! Feel free to add my:
>
> Tested-by: Guilherme G. Piccoli <[email protected]>
>

Reviewed-by: Matthias Brugger <[email protected]>

I gave that yesterday, but Weichen Chen seems to have forgotten about it.

Regards,
Matthias