Re: [PATCH] mm/vmalloc: Move free_vm_area(area) from the __vmalloc_area_node function to the __vmalloc_node_range_noprof function

[Uladzislau Rezki <urezki@gmail.com>]

On Thu, Mar 13, 2025 at 03:56:07PM +0530, Dev Jain wrote:
> 
> 
> On 13/03/25 2:48 pm, Lai, Yi wrote:
> > On Mon, Mar 03, 2025 at 09:57:02AM +0800, Liu Ye wrote:
> > > Moved free_vm_area from the __vmalloc_area_node function to the
> > > __vmalloc_node_range_noprof function so that allocation and freeing
> > > of the area can be paired in one function for better readability.
> > > 
> > > Signed-off-by: Liu Ye <liuye@kylinos.cn>
> > > ---
> > >   mm/vmalloc.c | 5 +++--
> > >   1 file changed, 3 insertions(+), 2 deletions(-)
> > > 
> > > diff --git a/mm/vmalloc.c b/mm/vmalloc.c
> > > index a6e7acebe9ad..dc658d4af181 100644
> > > --- a/mm/vmalloc.c
> > > +++ b/mm/vmalloc.c
> > > @@ -3651,7 +3651,6 @@ static void *__vmalloc_area_node(struct vm_struct *area, gfp_t gfp_mask,
> > >   		warn_alloc(gfp_mask, NULL,
> > >   			"vmalloc error: size %lu, failed to allocated page array size %lu",
> > >   			nr_small_pages * PAGE_SIZE, array_size);
> > > -		free_vm_area(area);
> > >   		return NULL;
> > >   	}
> > > @@ -3844,8 +3843,10 @@ void *__vmalloc_node_range_noprof(unsigned long size, unsigned long align,
> > >   	/* Allocate physical pages and map them into vmalloc space. */
> > >   	ret = __vmalloc_area_node(area, gfp_mask, prot, shift, node);
> > > -	if (!ret)
> > > +	if (!ret) {
> > > +		free_vm_area(area);
> > >   		goto fail;
> > > +	}
> > >   	/*
> > >   	 * Mark the pages as accessible, now that they are mapped.
> > > -- 
> > > 2.25.1
> > > 
> > 
> > Another issue is also bisected to this commit: there is KASAN: slab-use-after-free Read in io_create_region
> > 
> > 
> > Syzkaller repro code:
> > https://github.com/laifryiee/syzkaller_logs/tree/main/250313_124102_io_create_region/repro.c
> > Syzkaller repro syscall steps:
> > https://github.com/laifryiee/syzkaller_logs/tree/main/250313_124102_io_create_region/repro.prog
> > Syzkaller report:
> > https://github.com/laifryiee/syzkaller_logs/tree/main/250313_124102_io_create_region/repro.report
> > Issue dmesg:
> > https://github.com/laifryiee/syzkaller_logs/blob/main/250313_124102_io_create_region/eea255893718268e1ab852fb52f70c613d109b99_dmesg.log
> > 
> > "
> > [   24.707888]  __asan_report_load8_noabort+0x18/0x20
> > [   24.707895]  __vmalloc_node_range_noprof+0x1344/0x1420
> > [   24.707901]  ? __lock_acquire+0x1bad/0x5d60
> > [   24.707915]  ? io_create_region+0x6ca/0xe00
> > [   24.707933]  ? __pfx___vmalloc_node_range_noprof+0x10/0x10
> > [   24.707940]  ? debug_smp_processor_id+0x20/0x30
> > [   24.707958]  __kvmalloc_node_noprof+0x2a9/0x6a0
> > [   24.707969]  ? io_create_region+0x6ca/0xe00
> > [   24.707979]  ? io_create_region+0x6ca/0xe00
> > [   24.707992]  io_create_region+0x6ca/0xe00
> > [   24.708002]  ? io_create_region+0x6ca/0xe00
> > [   24.708015]  ? __pfx_io_create_region+0x10/0x10
> > [   24.708025]  ? __this_cpu_preempt_check+0x21/0x30
> > [   24.708035]  ? lock_release+0x440/0x870
> > [   24.708046]  ? __pfx_lock_release+0x10/0x10
> > [   24.708056]  io_create_region_mmap_safe+0xaf/0x150
> > [   24.708067]  ? __pfx_io_create_region_mmap_safe+0x10/0x10
> > [   24.708081]  ? __sanitizer_cov_trace_const_cmp8+0x1c/0x30
> > [   24.708096]  __do_sys_io_uring_register+0xb90/0x2b10
> > [   24.708110]  ? __pfx___do_sys_io_uring_register+0x10/0x10
> > [   24.708127]  ? trace_hardirqs_on+0x51/0x60
> > [   24.708145]  ? __audit_syscall_entry+0x39c/0x500
> > [   24.708157]  __x64_sys_io_uring_register+0x9f/0x100
> > [   24.708168]  ? syscall_trace_enter+0x14d/0x280
> > [   24.708179]  x64_sys_call+0x1eef/0x2150
> > [   24.708188]  do_syscall_64+0x6d/0x140
> > [   24.708199]  entry_SYSCALL_64_after_hwframe+0x76/0x7e
> > [   24.708206] RIP: 0033:0x7f9276c3ee5d
> > [   24.708217] Code: ff c3 66 2e 0f 1f 84 00 00 00 00 00 90 f3 0f 1e fa 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 93 af 1b 00 f7 d8 64 89 01 48
> > [   24.708224] RSP: 002b:00007ffd41ff7c18 EFLAGS: 00000202 ORIG_RAX: 00000000000001ab
> > [   24.708234] RAX: ffffffffffffffda RBX: 0000000000000000 RCX: 00007f9276c3ee5d
> > [   24.708238] RDX: 0000000020003300 RSI: 0000000000000022 RDI: 0000000000000003
> > [   24.708242] RBP: 00007ffd41ff7c30 R08: 00007ffd41ff7c30 R09: 00007ffd41ff7c30
> > [   24.708246] R10: 0000000000000001 R11: 0000000000000202 R12: 00007ffd41ff7d88
> > [   24.708250] R13: 00000000004017f5 R14: 0000000000403e08 R15: 00007f9276f77000
> > [   24.708260]  </TASK>
> > [   24.708262]
> > [   24.719961] Allocated by task 1208:
> > [   24.720136]  kasan_save_stack+0x2c/0x60
> > [   24.720336]  kasan_save_track+0x18/0x40
> > [   24.720529]  kasan_save_alloc_info+0x3c/0x50
> > [   24.720744]  __kasan_kmalloc+0x88/0xa0
> > [   24.720933]  __kmalloc_cache_node_noprof+0x1d0/0x470
> > [   24.721174]  __get_vm_area_node+0xf7/0x260
> > [   24.721380]  __vmalloc_node_range_noprof+0x25c/0x1420
> > [   24.721622]  __kvmalloc_node_noprof+0x2a9/0x6a0
> > [   24.721848]  io_create_region+0x6ca/0xe00
> > [   24.722049]  io_create_region_mmap_safe+0xaf/0x150
> > [   24.722291]  __do_sys_io_uring_register+0xb90/0x2b10
> > [   24.722539]  __x64_sys_io_uring_register+0x9f/0x100
> > [   24.722782]  x64_sys_call+0x1eef/0x2150
> > [   24.722976]  do_syscall_64+0x6d/0x140
> > [   24.723159]  entry_SYSCALL_64_after_hwframe+0x76/0x7e
> > [   24.723405]
> > [   24.723491] Freed by task 1208:
> > [   24.723648]  kasan_save_stack+0x2c/0x60
> > [   24.723841]  kasan_save_track+0x18/0x40
> > [   24.724035]  kasan_save_free_info+0x3f/0x60
> > [   24.724247]  __kasan_slab_free+0x3d/0x60
> > [   24.724450]  kfree+0x28d/0x440
> > [   24.724610]  vfree+0x2ce/0x910
> > [   24.724769]  __vmalloc_node_range_noprof+0xeca/0x1420
> > [   24.725017]  __kvmalloc_node_noprof+0x2a9/0x6a0
> > [   24.725240]  io_create_region+0x6ca/0xe00
> > [   24.725442]  io_create_region_mmap_safe+0xaf/0x150
> > [   24.725682]  __do_sys_io_uring_register+0xb90/0x2b10
> > [   24.725934]  __x64_sys_io_uring_register+0x9f/0x100
> > [   24.726177]  x64_sys_call+0x1eef/0x2150
> > [   24.726365]  do_syscall_64+0x6d/0x140
> > [   24.726549]  entry_SYSCALL_64_after_hwframe+0x76/0x7e
> > [   24.726796]
> > [   24.726882] The buggy address belongs to the object at ffff8880166bfb80
> > [   24.726882]  which belongs to the cache kmalloc-64 of size 64
> > [   24.727442] The buggy address is located 8 bytes inside of
> > [   24.727442]  freed 64-byte region [ffff8880166bfb80, ffff8880166bfbc0)
> > [   24.727990]
> > [   24.728078] The buggy address belongs to the physical page:
> > [   24.728344] page: refcount:0 mapcount:0 mapping:0000000000000000 index:0x0 pfn:0x166bf
> > [   24.728712] anon flags: 0xfffffc0000000(node=0|zone=1|lastcpupid=0x1fffff)
> > [   24.729034] page_type: f5(slab)
> > [   24.729196] raw: 000fffffc0000000 ffff88800d4418c0 0000000000000000 dead000000000001
> > [   24.729562] raw: 0000000000000000 0000000080200020 00000000f5000000 0000000000000000
> > [   24.729922] page dumped because: kasan: bad access detected
> > [   24.730189]
> > [   24.730274] Memory state around the buggy address:
> > [   24.730503]  ffff8880166bfa80: fa fb fb fb fb fb fb fb fc fc fc fc fc fc fc fc
> > [   24.730839]  ffff8880166bfb00: fa fb fb fb fb fb fb fb fc fc fc fc fc fc fc fc
> > [   24.731179] >ffff8880166bfb80: fa fb fb fb fb fb fb fb fc fc fc fc fc fc fc fc
> > [   24.731518]                       ^
> > [   24.731696]  ffff8880166bfc00: fa fb fb fb fb fb fb fb fc fc fc fc fc fc fc fc
> > [   24.732032]  ffff8880166bfc80: fa fb fb fb fb fb fb fb fc fc fc fc fc fc fc fc
> > [   24.732369] ==================================================================
> > [   24.732787] Disabling lock debugging due to kernel taint
> > [   24.734673] ------------[ cut here ]------------
> > [   24.734907] Trying to vfree() bad address (00000000ea48f0e7)
> > [   24.735229] WARNING: CPU: 0 PID: 1207 at mm/vmalloc.c:3231 remove_vm_area+0x1c2/0x220
> > [   24.735601] Modules linked in:
> > [   24.735760] CPU: 0 UID: 0 PID: 1207 Comm: repro Tainted: G    B               6.14.0-rc6-next-20250311-eea255893718 #1
> > [   24.736307] Tainted: [B]=BAD_PAGE
> > [   24.736477] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.16.0-0-gd239552ce722-prebuilt.qemu.org 04/01/2014
> > [   24.737002] RIP: 0010:remove_vm_area+0x1c2/0x220
> > [   24.737230] Code: c3 cc cc cc cc e8 7e e2 a4 ff 49 81 ee 00 10 00 00 e9 51 ff ff ff e8 6d e2 a4 ff 4c 89 e6 48 c7 c7 a0 eb fa 85 e8 ce 71 63 ff <0f> 0b 45 31 ed eb bb 45 31 ed eb b6 e8 ed d2 0f 00 e9 10 ff ff ff
> > [   24.738072] RSP: 0000:ffff88804dbc78b8 EFLAGS: 00010292
> > "
> > 
> > Hope this cound be insightful to you.
> > 
> > Regards,
> > Yi Lai
> > 
> 
> 
> I guess this trace is something else, butiIf __vmalloc_area_node() fails
> while allocating physical memory (after we have allocated the pages array),
> you call vfree(area->addr), and then the patch does a free_vm_area() again,
> causing a double free.
>
JFYI:

The patch, that moves free_vm_area() has been dropped due to degrade.

--
Uladzislau Rezki