Re: [PATCH v1 05/26] ACPICA: Fix NULL pointer dereference in acpi_ev_address_space_dispatch()

[Guenter Roeck <linux@roeck-us.net>]

Hi,

On 3/23/26 16:26, Dumbre, Saket wrote:
> Hi Guenter,
> 
> Yes, the experimental AI is spot-on in this case (although this is a very rare corner case path)! Internal buffer can stay uninitialized when field object is NULL (in case of loading ACPI tables from memory, doesn't target the PCC region though, if anything maybe just the PCCT table? The flow gets really complex to track each individual ACPI table), but otherwise, the other 2 fields shouldn't depend to be initialized on the existence of field object. The easiest fix would be to have a nested inner if (field_obj) {.....initialize ctx->internal_buffer....} and unlink the dependency on field_obj not being NULL on the remaining 2 ctx initializations. Thanks a lot for bringing this up and please feel free to submit a PR for ACPICA (and get the credit you deserve) which will then be pipelined into a patch for the Linux kernel when Rafael picks up the new ACPICA release commits! If you can deploy the experimental (or better yet, the release product) AI to find more bugs like this 
> and confirm after a human review if they are indeed real or not, that would help make ACPICA a lot better!
> 

My knowledge about ACPI code and ACPI patch submission rules is just about zero.
I'll be happy to submit patches into the Linux kernel, but I don't think I'll have
the time to learn how to submit patches into ACPI, sorry.

Guenter