Re: [PATCH] serial: 8250: fix null-ptr-deref in serial8250_start_tx()

From: Yang Yingliang
Date: Tue Jul 21 2020 - 08:18:33 EST


On 2020/7/21 19:54, Yang Yingliang wrote:

On 2020/7/21 18:48, Greg KH wrote:
On Tue, Jul 21, 2020 at 02:38:52PM +0000, Yang Yingliang wrote:
I got null-ptr-deref in serial8250_start_tx():

[ÂÂ 78.114630] Unable to handle kernel NULL pointer dereference at virtual address 0000000000000000
[ÂÂ 78.123778] Mem abort info:
[ÂÂ 78.126560]ÂÂ ESR = 0x86000007
[ÂÂ 78.129603]ÂÂ EC = 0x21: IABT (current EL), IL = 32 bits
[ÂÂ 78.134891]ÂÂ SET = 0, FnV = 0
[ÂÂ 78.137933]ÂÂ EA = 0, S1PTW = 0
[ÂÂ 78.141064] user pgtable: 64k pages, 48-bit VAs, pgdp=00000027d41a8600
[ÂÂ 78.147562] [0000000000000000] pgd=00000027893f0003, p4d=00000027893f0003, pud=00000027893f0003, pmd=00000027c9a20003, pte=0000000000000000
[ÂÂ 78.160029] Internal error: Oops: 86000007 [#1] SMP
[ÂÂ 78.164886] Modules linked in: sunrpc vfat fat aes_ce_blk crypto_simd cryptd aes_ce_cipher crct10dif_ce ghash_ce sha2_ce sha256_arm64 sha1_ce ses enclosure sg sbsa_gwdt ipmi_ssif spi_dw_mmio sch_fq_codel vhost_net tun vhost vhost_iotlb tap ip_tables ext4 mbcache jbd2 ahci hisi_sas_v3_hw libahci hisi_sas_main libsas hns3 scsi_transport_sas hclge libata megaraid_sas ipmi_si hnae3 ipmi_devintf ipmi_msghandler br_netfilter bridge stp llc nvme nvme_core xt_sctp sctp libcrc32c dm_mod nbd
[ÂÂ 78.207383] CPU: 11 PID: 23258 Comm: null-ptr Not tainted 5.8.0-rc6+ #48
[ÂÂ 78.214056] Hardware name: Huawei TaiShan 2280 V2/BC82AMDC, BIOS 2280-V2 CS V3.B210.01 03/12/2020
[ÂÂ 78.222888] pstate: 80400089 (Nzcv daIf +PAN -UAO BTYPE=--)
[ÂÂ 78.228435] pc : 0x0
[ÂÂ 78.230618] lr : serial8250_start_tx+0x160/0x260
[ÂÂ 78.235215] sp : ffff800062eefb80
[ÂÂ 78.238517] x29: ffff800062eefb80 x28: 0000000000000fff
[ÂÂ 78.243807] x27: ffff800062eefd80 x26: ffff202fd83b3000
[ÂÂ 78.249098] x25: ffff800062eefd80 x24: ffff202fd83b3000
[ÂÂ 78.254388] x23: ffff002fc5e50be8 x22: 0000000000000002
[ÂÂ 78.259679] x21: 0000000000000001 x20: 0000000000000000
[ÂÂ 78.264969] x19: ffffa688827eecc8 x18: 0000000000000000
[ÂÂ 78.270259] x17: 0000000000000000 x16: 0000000000000000
[ÂÂ 78.275550] x15: ffffa68881bc67a8 x14: 00000000000002e6
[ÂÂ 78.280841] x13: ffffa68881bc67a8 x12: 000000000000c539
[ÂÂ 78.286131] x11: d37a6f4de9bd37a7 x10: ffffa68881cccff0
[ÂÂ 78.291421] x9 : ffffa68881bc6000 x8 : ffffa688819daa88
[ÂÂ 78.296711] x7 : ffffa688822a0f20 x6 : ffffa688819e0000
[ÂÂ 78.302002] x5 : ffff800062eef9d0 x4 : ffffa68881e707a8
[ÂÂ 78.307292] x3 : 0000000000000000 x2 : 0000000000000002
[ÂÂ 78.312582] x1 : 0000000000000001 x0 : ffffa688827eecc8
[ÂÂ 78.317873] Call trace:
[ÂÂ 78.320312]Â 0x0
[ÂÂ 78.322147]Â __uart_start.isra.9+0x64/0x78
[ÂÂ 78.326229]Â uart_start+0xb8/0x1c8
[ÂÂ 78.329620]Â uart_flush_chars+0x24/0x30
[ÂÂ 78.333442]Â n_tty_receive_buf_common+0x7b0/0xc30
[ÂÂ 78.338128]Â n_tty_receive_buf+0x44/0x2c8
[ÂÂ 78.342122]Â tty_ioctl+0x348/0x11f8
[ÂÂ 78.345599]Â ksys_ioctl+0xd8/0xf8
[ÂÂ 78.348903]Â __arm64_sys_ioctl+0x2c/0xc8
[ÂÂ 78.352812]Â el0_svc_common.constprop.2+0x88/0x1b0
[ÂÂ 78.357583]Â do_el0_svc+0x44/0xd0
[ÂÂ 78.360887]Â el0_sync_handler+0x14c/0x1d0
[ÂÂ 78.364880]Â el0_sync+0x140/0x180
[ÂÂ 78.368185] Code: bad PC value

SERIAL_PORT_DFNS is not defined on each arch, if it's not defined,
serial8250_set_defaults() won't be called in serial8250_isa_init_ports(),
so the p->serial_in pointer won't be initialized, and it leads a null-ptr-deref.
Fix this problem by calling serial8250_set_defaults() after init uart port.

Signed-off-by: Yang Yingliang <yangyingliang@xxxxxxxxxx>
---
 drivers/tty/serial/8250/8250_core.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
Does this fix a specific commit, or has this issue always been present?
What has caused it to happen now that no one else has seen this?

I think it's always been present on the arch that not defined SERIAL_PORT_DFNS.

I got this on arm64 and here is the C reproducer:

// autogenerated by syzkaller (https://github.com/google/syzkaller)

#define _GNU_SOURCE

#include <endian.h>
#include <stdint.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <sys/syscall.h>
#include <sys/types.h>
#include <unistd.h>

#ifndef __NR_ioctl
#define __NR_ioctl 29
#endif
#ifndef __NR_mmap
#define __NR_mmap 222
#endif
#ifndef __NR_openat
#define __NR_openat 56
#endif

uint64_t r[1] = {0xffffffffffffffff};

int main(void)
{
ÂÂÂ syscall(__NR_mmap, 0x20000000ul, 0x1000000ul, 3ul, 0x32ul, -1, 0);
ÂÂÂ intptr_t res = 0;
ÂÂÂ memcpy((void*)0x20000040, "/dev/ttyS3\000", 11);
ÂÂÂ res = syscall(__NR_openat, 0xffffffffffffff9cul, 0x20000040ul, 0x401ul, 0ul);
ÂÂÂ if (res != -1)
ÂÂÂ ÂÂÂ r[0] = res;
ÂÂÂ syscall(__NR_ioctl, r[0], 0x5412ul, 0x20000080ul);
ÂÂÂ return 0;
}

It's can be reproduced with CONFIG_SERIAL_8250_RUNTIME_UARTS=8 and

by opening ttyS4 on x86_64. Because the size of SERIAL_PORT_DFNS is 4 and

the pointers of ttyS4 ~ ttyS7 won't be initialized.


[ÂÂ 49.427884] BUG: unable to handle kernel NULL pointer dereference at 0000000000000000
[ÂÂ 49.432786] PGD 80000003d7573067 P4D 80000003d7573067 PUD 3d7526067 PMD 0
[ÂÂ 49.436944] Oops: 0010 [#1] SMP KASAN PTI
[ÂÂ 49.439490] CPU: 1 PID: 2491 Comm: test Not tainted 4.19.90+ #1083
[ÂÂ 49.443437] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.13.0-0-gf21b5a4aeb02-prebuilt.qemu.org 04/01/2014
[ÂÂ 49.450233] RIP: 0010:ÂÂÂÂÂÂÂÂÂ (null)
[ÂÂ 49.452279] Code: Bad RIP value.
[ÂÂ 49.454066] RSP: 0018:ffff8883d73a7880 EFLAGS: 00010093
[ÂÂ 49.457082] RAX: ffff8883d72d6000 RBX: ffffffff85e243a0 RCX: ffffffff820a062b
[ÂÂ 49.460916] RDX: 0000000000000002 RSI: 0000000000000001 RDI: ffffffff85e243a0
[ÂÂ 49.464490] RBP: 0000000000000002 R08: fffffbfff0bc4875 R09: fffffbfff0bc4874
[ÂÂ 49.467993] R10: fffffbfff0bc4874 R11: ffffffff85e243a3 R12: ffffffff85e245f6
[ÂÂ 49.471526] R13: ffffffff85e243f0 R14: 0000000000000000 R15: 0000000000000000
[ÂÂ 49.475192] FS:Â 0000000002064880(0000) GS:ffff8883eb240000(0000) knlGS:0000000000000000
[ÂÂ 49.479019] CS:Â 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[ÂÂ 49.481852] CR2: ffffffffffffffd6 CR3: 00000003d77a2006 CR4: 0000000000020ee0
[ÂÂ 49.485131] Call Trace:
[ÂÂ 49.486278]Â serial8250_start_tx+0x56c/0x9b0
[ÂÂ 49.488167]Â __uart_start.isra.1+0x16a/0x1b0
[ÂÂ 49.490082]Â uart_start+0x1a7/0x340
[ÂÂ 49.491650]Â ? __uart_start.isra.1+0x1b0/0x1b0
[ÂÂ 49.493670]Â ? __process_echoes+0x636/0x860
[ÂÂ 49.495540]Â ? uart_start+0x340/0x340
[ÂÂ 49.497442]Â n_tty_receive_buf_common+0x10ff/0x2460
[ÂÂ 49.499897]Â ? n_tty_receive_buf2+0x40/0x40
[ÂÂ 49.501915]Â tty_ioctl+0x6e1/0x1400
[ÂÂ 49.503629]Â ? quarantine_put+0xb7/0x150
[ÂÂ 49.505474]Â ? tty_vhangup+0x30/0x30
[ÂÂ 49.507237]Â ? lockdep_hardirqs_on+0x290/0x560
[ÂÂ 49.509419]Â ? __kasan_slab_free+0x1ce/0x230
[ÂÂ 49.511490]Â ? kmem_cache_free+0xa6/0x2a0
[ÂÂ 49.513454]Â ? putname+0xd4/0x110
[ÂÂ 49.515059]Â ? do_sys_open+0x291/0x3f0
[ÂÂ 49.516888]Â ? do_syscall_64+0x11c/0xe08
[ÂÂ 49.518795]Â ? entry_SYSCALL_64_after_hwframe+0x49/0xbe
[ÂÂ 49.521359]Â ? __lock_acquire+0x5eb/0x2e50
[ÂÂ 49.523278]Â ? restore_nameidata+0x108/0x180
[ÂÂ 49.525242]Â ? tty_vhangup+0x30/0x30
[ÂÂ 49.526895]Â do_vfs_ioctl+0x1a5/0x1100
[ÂÂ 49.528608]Â ? may_open_dev+0xd0/0xd0
[ÂÂ 49.530414]Â ? ioctl_preallocate+0x1d0/0x1d0
[ÂÂ 49.532607]Â ? putname+0xd4/0x110
[ÂÂ 49.534097]Â ? rcu_read_lock_sched_held+0x114/0x130
[ÂÂ 49.536342]Â ? kmem_cache_free+0x263/0x2a0
[ÂÂ 49.538450]Â ? putname+0xd4/0x110
[ÂÂ 49.540036]Â ? do_sys_open+0x1ec/0x3f0
[ÂÂ 49.541909]Â ksys_ioctl+0x7c/0xa0
[ÂÂ 49.543458]Â __x64_sys_ioctl+0x74/0xb0
[ÂÂ 49.545433]Â ? lockdep_hardirqs_on+0x37c/0x560
[ÂÂ 49.547642]Â do_syscall_64+0x11c/0xe08
[ÂÂ 49.549447]Â ? syscall_return_slowpath+0x3a0/0x3a0
[ÂÂ 49.551741]Â ? trace_hardirqs_on_thunk+0x1a/0x1c
[ÂÂ 49.553810]Â ? trace_hardirqs_off_caller+0x54/0x190
[ÂÂ 49.556125]Â ? trace_hardirqs_off_thunk+0x1a/0x1c
[ÂÂ 49.558453]Â entry_SYSCALL_64_after_hwframe+0x49/0xbe


Thanks,

Yang


thanks,

greg k-h
.

.