[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [PATCH 7/9] hw/misc/aspeed_hace: Do not crash if address_space_map()
From: |
Peter Delevoryas |
Subject: |
Re: [PATCH 7/9] hw/misc/aspeed_hace: Do not crash if address_space_map() failed |
Date: |
Thu, 29 Dec 2022 12:54:36 -0800 |
On Thu, Dec 29, 2022 at 04:23:23PM +0100, Philippe Mathieu-Daudé wrote:
> address_space_map() can fail:
>
> uart:~$ hash test
> sha256_test
> tv[0]:
> Segmentation fault: 11
> Thread 3 "qemu-system-arm" received signal SIGSEGV, Segmentation fault.
> gen_acc_mode_iov (req_len=0x7ffff18b7778, id=<optimized out>,
> iov=0x7ffff18b7780, s=0x555556ce0bd0)
> at ../hw/misc/aspeed_hace.c:171
> 171 if (has_padding(s, &iov[id], *req_len, &total_msg_len,
> &pad_offset)) {
> (gdb) bt
> #0 gen_acc_mode_iov (req_len=0x7ffff18b7778, id=<optimized out>,
> iov=0x7ffff18b7780, s=0x555556ce0bd0)
> at ../hw/misc/aspeed_hace.c:171
> #1 do_hash_operation (s=s@entry=0x555556ce0bd0, algo=3,
> sg_mode=sg_mode@entry=true, acc_mode=acc_mode@entry=true)
> at ../hw/misc/aspeed_hace.c:224
> #2 0x00005555559bdbb8 in aspeed_hace_write (opaque=<optimized out>,
> addr=12, data=262488, size=<optimized out>)
> at ../hw/misc/aspeed_hace.c:358
>
> This change doesn't fix much, but at least the guest
> can't crash QEMU anymore. Instead it is still usable:
>
> uart:~$ hash test
> sha256_test
> tv[0]:hash_final error
> sha384_test
> tv[0]:hash_final error
> sha512_test
> tv[0]:hash_final error
> [00:00:06.278,000] <err> hace_global: HACE poll timeout
> [00:00:09.324,000] <err> hace_global: HACE poll timeout
> [00:00:12.261,000] <err> hace_global: HACE poll timeout
> uart:~$
>
> Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Good catch,
Reviewed-by: Peter Delevoryas <peter@pjd.dev>
Maybe include this tag?
Fixes: c5475b3f9a ("hw: Model ASPEED's Hash and Crypto Engine")
> ---
> hw/misc/aspeed_hace.c | 21 +++++++++++++++------
> 1 file changed, 15 insertions(+), 6 deletions(-)
>
> diff --git a/hw/misc/aspeed_hace.c b/hw/misc/aspeed_hace.c
> index ac21be306c..12a761f1f5 100644
> --- a/hw/misc/aspeed_hace.c
> +++ b/hw/misc/aspeed_hace.c
> @@ -193,6 +193,7 @@ static void do_hash_operation(AspeedHACEState *s, int
> algo, bool sg_mode,
> size_t digest_len = 0;
> int niov = 0;
> int i;
> + void *haddr;
>
> if (sg_mode) {
> uint32_t len = 0;
> @@ -217,9 +218,13 @@ static void do_hash_operation(AspeedHACEState *s, int
> algo, bool sg_mode,
> addr &= SG_LIST_ADDR_MASK;
>
> plen = len & SG_LIST_LEN_MASK;
> - iov[i].iov_base = address_space_map(&s->dram_as, addr, &plen,
> false,
> - MEMTXATTRS_UNSPECIFIED);
> -
> + haddr = address_space_map(&s->dram_as, addr, &plen, false,
> + MEMTXATTRS_UNSPECIFIED);
> + if (haddr == NULL) {
> + qemu_log_mask(LOG_GUEST_ERROR, "%s: qcrypto failed\n",
> __func__);
> + return;
> + }
> + iov[i].iov_base = haddr;
> if (acc_mode) {
> niov = gen_acc_mode_iov(s, iov, i, &plen);
>
> @@ -230,10 +235,14 @@ static void do_hash_operation(AspeedHACEState *s, int
> algo, bool sg_mode,
> } else {
> hwaddr len = s->regs[R_HASH_SRC_LEN];
>
> + haddr = address_space_map(&s->dram_as, s->regs[R_HASH_SRC],
> + &len, false, MEMTXATTRS_UNSPECIFIED);
> + if (haddr == NULL) {
> + qemu_log_mask(LOG_GUEST_ERROR, "%s: qcrypto failed\n", __func__);
> + return;
> + }
> + iov[0].iov_base = haddr;
> iov[0].iov_len = len;
> - iov[0].iov_base = address_space_map(&s->dram_as, s->regs[R_HASH_SRC],
> - &len, false,
> - MEMTXATTRS_UNSPECIFIED);
> i = 1;
>
> if (s->iov_count) {
> --
> 2.38.1
>
- Re: [PATCH 3/9] hw/arm/aspeed_ast10x0: Add various unimplemented peripherals, (continued)
- [PATCH 4/9] hw/arm/aspeed_ast10x0: Map I3C peripheral, Philippe Mathieu-Daudé, 2022/12/29
- [PATCH 5/9] hw/arm/aspeed_ast10x0: Map the secure SRAM, Philippe Mathieu-Daudé, 2022/12/29
- [PATCH 6/9] hw/arm/aspeed_ast10x0: Map HACE peripheral, Philippe Mathieu-Daudé, 2022/12/29
- [PATCH 7/9] hw/misc/aspeed_hace: Do not crash if address_space_map() failed, Philippe Mathieu-Daudé, 2022/12/29
- Re: [PATCH 7/9] hw/misc/aspeed_hace: Do not crash if address_space_map() failed,
Peter Delevoryas <=
- [PATCH 8/9] hw/arm/aspeed_ast10x0: Add TODO comment to use Cortex-M4F, Philippe Mathieu-Daudé, 2022/12/29
- [PATCH 9/9] tests/avocado: Test Aspeed Zephyr SDK v00.01.08 on AST1030 board, Philippe Mathieu-Daudé, 2022/12/29
- Re: [PATCH 0/9] hw/arm/aspeed_ast10x0: Map more peripherals & few more fixes, Peter Delevoryas, 2022/12/29