On 24-02-27 00:53, Andrey Skvortsov wrote:
> sun8i_ce_cipher_unprepare should be called before
> crypto_finalize_skcipher_request, because client callbacks may
> immediately free memory, that isn't needed anymore. But it will be
> used by unprepare after free. Before removing prepare/unprepare
> callbacks it was handled by crypto engine in crypto_finalize_request.
And potentially rk3288_crypto driver is affected by the similar
problem.
https://elixir.bootlin.com/linux/v6.8-rc6/source/drivers/crypto/rockchip/rk3288_crypto_ahash.c#L339
>
> Usually that results in a pointer dereference problem during a in
> crypto selftest.
> Unable to handle kernel NULL pointer dereference at
> virtual address 0000000000000030
> Mem abort info:
> ESR = 0x0000000096000004
> EC = 0x25: DABT (current EL), IL = 32 bits
> SET = 0, FnV = 0
> EA = 0, S1PTW = 0
> FSC = 0x04: level 0 translation fault
> Data abort info:
> ISV = 0, ISS = 0x00000004, ISS2 = 0x00000000
> CM = 0, WnR = 0, TnD = 0, TagAccess = 0
> GCS = 0, Overlay = 0, DirtyBit = 0, Xs = 0
> user pgtable: 4k pages, 48-bit VAs, pgdp=000000004716d000
> [0000000000000030] pgd=0000000000000000, p4d=0000000000000000
> Internal error: Oops: 0000000096000004 [#1] SMP
>
> This problem is detected by KASAN as well.
> ==================================================================
> BUG: KASAN: slab-use-after-free in sun8i_ce_cipher_do_one+0x6e8/0xf80 [sun8i_ce]
> Read of size 8 at addr ffff00000dcdc040 by task 1c15000.crypto-/373
>
> Hardware name: Pine64 PinePhone (1.2) (DT)
> Call trace:
> dump_backtrace+0x9c/0x128
> show_stack+0x20/0x38
> dump_stack_lvl+0x48/0x60
> print_report+0xf8/0x5d8
> kasan_report+0x90/0xd0
> __asan_load8+0x9c/0xc0
> sun8i_ce_cipher_do_one+0x6e8/0xf80 [sun8i_ce]
> crypto_pump_work+0x354/0x620 [crypto_engine]
> kthread_worker_fn+0x244/0x498
> kthread+0x168/0x178
> ret_from_fork+0x10/0x20
>
> Allocated by task 379:
> kasan_save_stack+0x3c/0x68
> kasan_set_track+0x2c/0x40
> kasan_save_alloc_info+0x24/0x38
> __kasan_kmalloc+0xd4/0xd8
> __kmalloc+0x74/0x1d0
> alg_test_skcipher+0x90/0x1f0
> alg_test+0x24c/0x830
> cryptomgr_test+0x38/0x60
> kthread+0x168/0x178
> ret_from_fork+0x10/0x20
>
> Freed by task 379:
> kasan_save_stack+0x3c/0x68
> kasan_set_track+0x2c/0x40
> kasan_save_free_info+0x38/0x60
> __kasan_slab_free+0x100/0x170
> slab_free_freelist_hook+0xd4/0x1e8
> __kmem_cache_free+0x15c/0x290
> kfree+0x74/0x100
> kfree_sensitive+0x80/0xb0
> alg_test_skcipher+0x12c/0x1f0
> alg_test+0x24c/0x830
> cryptomgr_test+0x38/0x60
> kthread+0x168/0x178
> ret_from_fork+0x10/0x20
>
> The buggy address belongs to the object at ffff00000dcdc000
> which belongs to the cache kmalloc-256 of size 256
> The buggy address is located 64 bytes inside of
> freed 256-byte region [ffff00000dcdc000, ffff00000dcdc100)
>
> Signed-off-by: Andrey Skvortsov <[email protected]>
> Fixes: 4136212ab18e ("crypto: sun8i-ce - Remove prepare/unprepare request")
--
Best regards,
Andrey Skvortsov
On Tue, Feb 27, 2024 at 01:05:30AM +0300, Andrey Skvortsov wrote:
>
> And potentially rk3288_crypto driver is affected by the similar
> problem.
Indeed, and there is one more in sun8i-ce. I'll send out patches
for them.
Thanks,
--
Email: Herbert Xu <[email protected]>
Home Page: http://gondor.apana.org.au/~herbert/
PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt