[PATCH] setupapi: Avoid race conditions with devnode_table.

[Zebediah Figura (she/her)]

On 2/8/22 13:33, Brendan Shanks wrote:
> Signed-off-by: Brendan Shanks <bshanks at codeweavers.com>
> ---
> 
> Fixes a crash where two threads calling SetupDiGetClassDevs() would end
> up in alloc_devnode() at the same time. One thread would start iterating
> through the table after devnode_table_size was set but before
> devnode_table was allocated, and crash.

If we need setupapi to be thread-safe, does it make sense just to put a 
lock around all devinst functions instead?