[PATCH v3] arm64/mm: avoid fixmap race condition when create pud mapping

[Catalin Marinas]

On Thu, Dec 16, 2021 at 04:28:12PM +0800, Jianyong Wu wrote:
> The 'fixmap' is a global resource and is used recursively by
> create pud mapping(), leading to a potential race condition in the
> presence of a concurrent call to alloc_init_pud():
> 
> kernel_init thread                          virtio-mem workqueue thread
> ==================                          ===========================
> 
>   alloc_init_pud(...)                       alloc_init_pud(...)
>   pudp = pud_set_fixmap_offset(...)         pudp = pud_set_fixmap_offset(...)
>   READ_ONCE(*pudp)
>   pud_clear_fixmap(...)
>                                             READ_ONCE(*pudp) // CRASH!
> 
> As kernel may sleep during creating pud mapping, introduce a mutex lock to
> serialise use of the fixmap entries by alloc_init_pud().
> 
> Signed-off-by: Jianyong Wu <jianyong.wu at arm.com>

I tried to queue this patch but with certain configurations it doesn't
boot under Qemu. Starting from defconfig, update .config with (I had
this in one of my build scripts):

$ ./scripts/config \
		-e DEBUG_KERNEL \
		-e DEBUG_PAGEALLOC \
		-e DEBUG_PAGEALLOC_ENABLE_DEFAULT \
		-e DEBUG_WX \
		-e DEBUG_SET_MODULE_RONX \
		-e DEBUG_ALIGN_RODATA \
		-e ARM64_PTDUMP_DEBUGFS \
		-e DEBUG_OBJECTS \
		-e DEBUG_OBJECTS_FREE \
		-e DEBUG_OBJECTS_TIMERS \
		-e DEBUG_KOBJECT_RELEASE \
		-e DEBUG_LOCKING_API_SELFTESTS \
		-e DEBUG_PREEMPT \
		-e DEBUG_TIMEKEEPING \
		-e DEBUG_VM \
		-e DEBUG_VM_VMACACHE \
		-e DEBUG_VM_RB \
		-e DEBUG_VM_PGFLAGS \
		-e DEBUG_VIRTUAL \
		-e DEBUG_LIST \
		-e DEBUG_PI_LIST \
		-e DEBUG_SG \
		-e PROVE_LOCKING \
		-e DEBUG_RT_MUTEXES \
		-e DEBUG_ATOMIC_SLEEP \
		-e ATOMIC64_SELFTEST

It stop after exiting the EFI boot services. I did not have time to
debug.

-- 
Catalin