On 03/03/2021 23:08, Gedare Bloom wrote:
I think it is just invalid to shrink the stack size smaller than it is
requested.
If I call pthread_attr_setstacksize(), then "The stacksize attribute
shall define the minimum stack size (in bytes) allocated for the
created threads stack." I should not get back a stack size smaller
than I requested when I create the pthread. It's just wrong.
https://pubs.opengroup.org/onlinepubs/9699919799/functions/pthread_attr_getstacksize.html
Allocated is one thing, usable is another thing. Stacks are
architecture-dependent. For example on powerpc, you have to set up an
initial stack frame at the top of the stack:
void _CPU_Context_Initialize(
Context_Control *the_context,
void *stack_base,
size_t size,
uint32_t new_level,
void *entry_point,
bool is_fp,
void *tls_area
)
{
ppc_context *the_ppc_context;
uint32_t msr_value = 0;
uintptr_t sp;
uintptr_t stack_alignment;
sp = (uintptr_t) stack_base + size - PPC_MINIMUM_STACK_FRAME_SIZE;
stack_alignment = CPU_STACK_ALIGNMENT;
sp &= ~(stack_alignment - 1);
sp = (uintptr_t) memset((void *) sp, 0, PPC_MINIMUM_STACK_FRAME_SIZE);
Here we align the top of stack and set the initial frame.
I am not sure which actual problem we have to solve currently. Is this
related to your CANARY example? In order to support things like this, we
probably need an API to get the usable/actual/whatever stack area for a
thread.
If you want to enforce that a user-provided size is the minimum
allocated size, then I think we have to align it up to the next stack
boundary. This makes sense from my point of view, however, this is not
the purpose of the current patch, it addresses issues related to the
heap allocator vs. the required stack alignment.
--
embedded brains GmbH
Herr Sebastian HUBER
Dornierstr. 4
82178 Puchheim
Germany
email: sebastian.hu...@embedded-brains.de
phone: +49-89-18 94 741 - 16
fax: +49-89-18 94 741 - 08
Registergericht: Amtsgericht München
Registernummer: HRB 157899
Vertretungsberechtigte Geschäftsführer: Peter Rasmussen, Thomas Dörfler
Unsere Datenschutzerklärung finden Sie hier:
https://embedded-brains.de/datenschutzerklaerung/
_______________________________________________
devel mailing list
devel@rtems.org
http://lists.rtems.org/mailman/listinfo/devel
