linux-hardening - Re: [RFC PATCH 2/6] powerpc/64s: Helpers to switch between linear and vmapped stack pointers

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Sat, 5 Nov 2022 20:28:27 +0100
From:   Christophe Leroy <christophe.leroy@...roup.eu>
To:     Andrew Donnellan <ajd@...ux.ibm.com>,
        "linuxppc-dev@...ts.ozlabs.org" <linuxppc-dev@...ts.ozlabs.org>
Cc:     "linux-hardening@...r.kernel.org" <linux-hardening@...r.kernel.org>,
        "cmr@...escreens.de" <cmr@...escreens.de>,
        Nick Piggin <npiggin@...il.com>
Subject: Re: [RFC PATCH 2/6] powerpc/64s: Helpers to switch between linear and
 vmapped stack pointers



Le 05/11/2022 à 09:00, Christophe Leroy a écrit :
> 
> 
> Le 04/11/2022 à 18:27, Andrew Donnellan a écrit :
>> powerpc unfortunately has too many places where we run stuff in real mode.
>>
>> With CONFIG_VMAP_STACK enabled, this means we need to be able to swap the
>> stack pointer to use the linear mapping when we enter a real mode section,
>> and back afterwards.
>>
>> Store the top bits of the stack pointer in both the linear map and the
>> vmalloc space in the PACA, and add some helper macros/functions to swap
>> between them.
> 
> That may work when pagesize is 64k because stack is on a single page,
> but I doubt is works with 4k pages, because vmalloc may allocate non
> contiguous pages.
> 

[snip]

> 
>> +
>> +#else // __ASSEMBLY__
>> +
>> +#include <asm/paca.h>
>> +#include <asm/reg.h>
>> +#include <linux/mm.h>
>> +
>> +#define stack_pa(ptr) (is_vmalloc_addr((ptr)) ? (void *)vmalloc_to_phys((void *)(ptr)) : (void *)ptr)
>> +
>> +static __always_inline void swap_stack_linear(void)
>> +{
>> +	current_stack_pointer = get_paca()->kstack_linear_base |	\
>> +		(current_stack_pointer & (THREAD_SIZE - 1));
> 
> That looks hacky. I think you can't just change current_stack_pointer on
> the fly. You have to provide something similar to call_do_softirq() or
> call_do_irq()
> 

Maybe you can have a look at Nic's RFC for calling functions in real 
mode : 
https://patchwork.ozlabs.org/project/linuxppc-dev/patch/20210212012041.392566-1-npiggin@gmail.com/

Christophe