RE: [PATCH] platform/x86: dell-smbios-base: Fix use after free on failure of dell_smbios_init()

From: Mario.Limonciello
Date: Wed Apr 03 2019 - 21:31:47 EST

Next message: Steven Rostedt: "Re: [PATCH V32 19/27] x86/mmiotrace: Lock down the testmmiotrace module"
Previous message: Anson Huang: "RE: [PATCH 1/2] input: keyboard: imx: no need to control interrupt status in event check"
In reply to: Darren Hart: "Re: [PATCH] platform/x86: dell-smbios-base: Fix use after free on failure of dell_smbios_init()"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

> -----Original Message-----
> From: Darren Hart <dvhart@xxxxxxxxxxxxx>
> Sent: Wednesday, April 3, 2019 3:16 PM
> To: Steven Rostedt; Limonciello, Mario
> Cc: LKML; platform-driver-x86@xxxxxxxxxxxxxxx; Andy Shevchenko; Pali Rohár; Tom
> Zanussi
> Subject: Re: [PATCH] platform/x86: dell-smbios-base: Fix use after free on failure of
> dell_smbios_init()
>
>
> [EXTERNAL EMAIL]
>
> On Wed, Apr 03, 2019 at 03:20:18PM -0400, Steven Rostedt wrote:
> > From: "Steven Rostedt (VMware)" <rostedt@xxxxxxxxxxx>
> >
> > If da_tokens are allocated, but dell_smbios_init() eventually fails, it will
> > free the da_tokens but it does not reset the da_num_tokens number. This
> > leads to the possibility of a use after free in dell_smbios_find_token().
> > As da_tokens is not NULL and da_num_tokens is set to something other than 0.
> >
> > By reseting the da_num_tokens to zero, and da_tokens to NULL after it is
> > freed, then access into the other functions that reference them will not
> > read freed memory.
>
> Upon closer inspections this appears to be a race more than an access
> issue. Even with this patch, there is still the space between kfree()
> and da_num_tokens = 0. This could be addressed by setting da_num_tokens
> to 0 prior to the call to kfree().
>
> That said, the bigger issue here seems to be the ability to use the
> dell_smbios_find_token() call before the module_init() has completed
> successfully.
>
> Mario, care to weigh in?
>

As I can tell as soon as da_num_tokens is set to 0 that function should be safe to call
though even if module_init wasn't finished yet.

> >
> > This was caught by a KASAN report:
> >
> > ==================================================================
> > BUG: KASAN: use-after-free in dell_smbios_find_token+0x2e/0x80 [dell_smbios]
> > Read of size 2 at addr ffff88840c2bc1a8 by task systemd-udevd/479
> >
> > CPU: 0 PID: 479 Comm: systemd-udevd Not tainted 5.1.0-rc1+ #9
> > Hardware name: Dell Inc. XPS 13 9360/02PG84, BIOS 2.3.1 10/03/2017
> > Call Trace:
> > dump_stack+0x7c/0xbb
> > ? dell_smbios_find_token+0x2e/0x80 [dell_smbios]
> > print_address_description+0xc7/0x280
> > ? dell_smbios_find_token+0x2e/0x80 [dell_smbios]
> > ? dell_smbios_find_token+0x2e/0x80 [dell_smbios]
> > kasan_report+0x14e/0x192
> > ? dell_smbios_find_token+0x2e/0x80 [dell_smbios]
> > dell_smbios_find_token+0x2e/0x80 [dell_smbios]
> > kbd_led_init+0x2e7/0x473 [dell_laptop]
> > ? dmi_matched+0x2a/0x2a [dell_laptop]
> > ? get_device_parent.isra.28+0x2a0/0x2a0
> > ? lockdep_init_map+0x98/0x2c0
> > ? platform_device_add+0x1b5/0x3a0
> > dell_init+0x4ad/0xb63 [dell_laptop]
> > ? kbd_led_init+0x473/0x473 [dell_laptop]
> > ? ___slab_alloc+0x61f/0x700
> > ? ___slab_alloc+0x61f/0x700
> > ? preempt_count_sub+0x15/0x100
> > ? kbd_led_init+0x473/0x473 [dell_laptop]
> > do_one_initcall+0xbd/0x3fd
> > ? perf_trace_initcall_level+0x280/0x280
> > ? kasan_unpoison_shadow+0x30/0x40
> > ? __kasan_kmalloc.constprop.8+0xa0/0xd0
> > ? kmem_cache_alloc_trace+0x163/0x390
> > ? kasan_unpoison_shadow+0x30/0x40
> > do_init_module+0xe3/0x341
> > load_module+0x2fc5/0x3ad0
> > ? layout_and_allocate+0x1170/0x1170
> > ? vfs_read+0xd4/0x1b0
> > ? kernel_read+0x74/0xa0
> > ? kernel_read_file+0x148/0x320
> > ? seccomp_notify_release+0x110/0x110
> > ? __do_sys_finit_module+0x192/0x1c0
> > __do_sys_finit_module+0x192/0x1c0
> > ? __ia32_sys_init_module+0x40/0x40
> > ? syscall_trace_enter+0x184/0x5e0
> > ? mark_held_locks+0x1a/0x90
> > do_syscall_64+0x72/0x220
> > entry_SYSCALL_64_after_hwframe+0x49/0xbe
> > RIP: 0033:0x7fcb4f5f5a49
> > Code: 00 f3 c3 66 2e 0f 1f 84 00 00 00 00 00 0f 1f 40 00 48 89 f8 48 89 f7
> > 48 89 d6 48 89 ca 4d 89 c2 4d 89 c8 4c 8b 4c 24 08 0f 05 <48> 3d 01 f0 ff ff
> > 73 01 c3 48 8b 0d 0f b4 2c 00 f7 d8 64 89 01 48
> > RSP: 002b:00007ffc73e340b8 EFLAGS: 00000246 ORIG_RAX: 0000000000000139
> > RAX: ffffffffffffffda RBX: 00005599992bb850 RCX: 00007fcb4f5f5a49
> > RDX: 0000000000000000 RSI: 00007fcb4f2e11c5 RDI: 0000000000000010
> > RBP: 00007fcb4f2e11c5 R08: 0000000000000000 R09: 00005599992bb850
> > R10: 0000000000000010 R11: 0000000000000246 R12: 0000000000000000
> > R13: 0000559999298f40 R14: 0000000000020000 R15: 00005599992bb850
> >
> > Allocated by task 479:
> > __kasan_kmalloc.constprop.8+0xa0/0xd0
> > krealloc+0xa0/0xc0
> > 0xffffffffc0cc0075
> > dmi_decode_table+0xf6/0x140
> > dmi_walk+0x46/0x70
> > 0xffffffffc0cc0109
> > do_one_initcall+0xbd/0x3fd
> > do_init_module+0xe3/0x341
> > load_module+0x2fc5/0x3ad0
> > __do_sys_finit_module+0x192/0x1c0
> > do_syscall_64+0x72/0x220
> > entry_SYSCALL_64_after_hwframe+0x49/0xbe
> >
> > Freed by task 479:
> > __kasan_slab_free+0x111/0x150
> > kfree+0xf5/0x350
> > 0xffffffffc0cc01d4
> > do_one_initcall+0xbd/0x3fd
> > do_init_module+0xe3/0x341
> > load_module+0x2fc5/0x3ad0
> > __do_sys_finit_module+0x192/0x1c0
> > do_syscall_64+0x72/0x220
> > entry_SYSCALL_64_after_hwframe+0x49/0xbe
> >
> > The buggy address belongs to the object at ffff88840c2bc1a8
> > which belongs to the cache kmalloc-2k of size 2048
> > The buggy address is located 0 bytes inside of
> > 2048-byte region [ffff88840c2bc1a8, ffff88840c2bc9a8)
> > The buggy address belongs to the page:
> > page:ffffea001030ae00 count:1 mapcount:0 mapping:ffff8884204113c0
> index:0x0
> > compound_mapcount: 0
> > flags: 0x17ffffc0010200(slab|head)
> > raw: 0017ffffc0010200 ffffea0010367608 ffffea000ea31808 ffff8884204113c0
> > raw: 0000000000000000 00000000000d000d 00000001ffffffff
> 0000000000000000
> > page dumped because: kasan: bad access detected
> >
> > Memory state around the buggy address:
> > ffff88840c2bc080: fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc
> > ffff88840c2bc100: fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc
> > >ffff88840c2bc180: fc fc fc fc fc fb fb fb fb fb fb fb fb fb fb fb
> > ^
> > ffff88840c2bc200: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
> > ffff88840c2bc280: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
> > ==================================================================
> >
> > Link: http://lkml.kernel.org/r/1553106560.2080.5.camel@xxxxxxxxx
> >
> > Reported-by: Tom Zanussi <tzanussi@xxxxxxxxx>
> > Tested-by: Tom Zanussi <tzanussi@xxxxxxxxx>
> > Signed-off-by: Steven Rostedt (VMware) <rostedt@xxxxxxxxxxx>
> > ---
> > drivers/platform/x86/dell-smbios-base.c | 2 ++
> > 1 file changed, 2 insertions(+)
> >
> > diff --git a/drivers/platform/x86/dell-smbios-base.c b/drivers/platform/x86/dell-
> smbios-base.c
> > index 0537d44d45a6..a74c0df25b15 100644
> > --- a/drivers/platform/x86/dell-smbios-base.c
> > +++ b/drivers/platform/x86/dell-smbios-base.c
> > @@ -625,6 +625,8 @@ static int __init dell_smbios_init(void)
> >
> > fail_platform_driver:
> > kfree(da_tokens);
> > + da_tokens = NULL;
> > + da_num_tokens = 0;
> > return ret;
> > }
> >
> > --
> > 2.20.1
> >
> >
>
> --
> Darren Hart
> VMware Open Source Technology Center

Next message: Steven Rostedt: "Re: [PATCH V32 19/27] x86/mmiotrace: Lock down the testmmiotrace module"
Previous message: Anson Huang: "RE: [PATCH 1/2] input: keyboard: imx: no need to control interrupt status in event check"
In reply to: Darren Hart: "Re: [PATCH] platform/x86: dell-smbios-base: Fix use after free on failure of dell_smbios_init()"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]