linux-kernel - [PATCH] mm: fix overflow in alloc_vmap

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [thread-next>] [day] [month] [year] [list]
Date:	Thu, 20 Jun 2013 18:12:30 +0300
From:	Ghennadi Procopciuc <unix140@...il.com>
To:	akpm@...ux-foundation.org, js1304@...il.com, rientjes@...gle.com,
	minchan@...nel.org, kosaki.motohiro@...fujitsu.com,
	linux-mm@...ck.org, linux-kernel@...r.kernel.org
Cc:	Ghennadi Procopciuc <unix140@...il.com>,
	Daniel Baluta <dbaluta@...acom.com>
Subject: [PATCH] mm: fix overflow in alloc_vmap_area

Inserting the following kernel module:

<snip>

static int simple_test_init(void)
{
        size_t i, j;
        void *address;

        for (i = 0 * MB; i<  60 * MB; i += 1 * MB) {
                for (j = i; j<  i + 1 * MB; j += KB) {
                        address = vmalloc(j);
                        vfree(address);
                }
        }

        return 0;
}

</snip>

triggers BUG at mm/vmalloc.c:310 on a x86 machine:

[   95.218283] Kernel BUG at c1126cdb [verbose debug info unavailable]
[   95.218306] invalid opcode: 0000 [#1] SMP
[   95.218324] Modules linked in: lkma_test(OF+)<snip lots of not tainted modules>
[   95.218559] Pid: 2419, comm: insmod Tainted: GF          O 3.9.0+ #57 Hewlett-Packard HP Compaq 8200 Elite CMT PC/1494
[   95.218597] EIP: 0060:[<c1126cdb>] EFLAGS: 00010207 CPU: 3
[   95.218617] EIP is at __insert_vmap_area+0xfb/0x100
[   95.218635] EAX: f85dc000 EBX: ef05cac0 ECX: f7be08c4 EDX: 00000007
[   95.218657] ESI: f2ed044c EDI: c1a220b8 EBP: f027bd34 ESP: f027bd14
[   95.218680]  DS: 007b ES: 007b FS: 00d8 GS: 00e0 SS: 0068
[   95.218699] CR0: 80050033 CR2: b5995118 CR3: 30364000 CR4: 000407f0
[   95.218721] DR0: 00000000 DR1: 00000000 DR2: 00000000 DR3: 00000000
[   95.218743] DR6: ffff0ff0 DR7: 00000400
[   95.218758] Process insmod (pid: 2419, ti=f027a000 task=efe64010 task.ti=f027a000)
[   95.218784] Stack:
[   95.218792]  c177b964 fecfb000 f85e2000 00000000 f85dc000 ffbfe000 f02da1c0 0007f000
[   95.218829]  f027bd7c c112802d c177b9b0 fecfb000 01305000 ffbfe000 00000001 ef05cac0
[   95.218866]  00000000 00000000 f83fe000 01304fff f83fe000 ffffffff 01305000 ef05c500
[   95.218903] Call Trace:
[   95.218915]  [<f85e2000>] ? 0xf85e1fff
[   95.218930]  [<f85dc000>] ? 0xf85dbfff
[   95.218945]  [<c112802d>] alloc_vmap_area.isra.16+0x1bd/0x2f0
[   95.218967]  [<c112867f>] __get_vm_area_node.isra.17+0x8f/0x130
[   95.218988]  [<c1128d77>] __vmalloc_node_range+0x57/0x200
[   95.219009]  [<f85f3075>] ? lkma_test_init+0x45/0x70 [lkma_test]
[   95.219031]  [<c1128f82>] __vmalloc_node+0x62/0x70
[   95.219049]  [<f85f3075>] ? lkma_test_init+0x45/0x70 [lkma_test]
[   95.219071]  [<c1129058>] vmalloc+0x38/0x40
[   95.219087]  [<f85f3075>] ? lkma_test_init+0x45/0x70 [lkma_test]
[   95.219109]  [<f85f3075>] lkma_test_init+0x45/0x70 [lkma_test]
[   95.219131]  [<f85f3030>] ? kzalloc+0x10/0x10 [lkma_test]
[   95.219151]  [<c1001222>] do_one_initcall+0x112/0x160
[   95.219171]  [<c15ca3cf>] ? set_section_ro_nx+0x54/0x59
[   95.219190]  [<c1099b69>] load_module+0x1d79/0x2660
[   95.219209]  [<c114721d>] ? create_object+0x19d/0x280
[   95.219230]  [<c109a4c8>] sys_init_module+0x78/0xb0
[   95.219250]  [<c15d9801>] sysenter_do_call+0x12/0x22
[   95.219268] Code: 39 03 73 0c 8b 3f 89 f0 83 c7 08 e9 3d ff ff ff 8b 46 f4 39 43 04 76 13 8b 3f 89 f0 83 c7 04 e9 29 ff ff ff e8 fb 1b 4a 00 eb ab<0f>  0b 8d 76 00 55 89 e5 56 53 66 66 66 66 90 83 60 0c df 89 c6
[   95.219415] EIP: [<c1126cdb>] __insert_vmap_area+0xfb/0x100 SS:ESP 0068:f027bd14
[   95.228313] ---[ end trace e0a1efb2acb97c98 ]---

A printk placed in __insert_vmap_area will show:

[   95.218256] va->va_start=0xfecfb000 tmp_va->va_end=0xf85e2000 va->va_end=0 tmp_va->va_start=0xf85dc000

and another one, before sum operation in alloc_vmap_area:

[   95.218204] addr = 0xfecfb000 size = 19943424 vend = 0xffbfe000

If after addition the result is smaller than one of the arguments,
then an overflow occurred. In our case there is an obvious overflow.

Signed-off-by: Ghennadi Procopciuc <unix140@...il.com>
Cc: Daniel Baluta<dbaluta@...acom.com>

---
 Don't know if this is the right solution, but the bug happens for me in
 3.10-rc6 and 3.9.
---
 mm/vmalloc.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/mm/vmalloc.c b/mm/vmalloc.c
index 0f751f2..6ea525c 100644
--- a/mm/vmalloc.c
+++ b/mm/vmalloc.c
@@ -421,12 +421,13 @@ nocache:
 	}

 found:
-	if (addr + size > vend)
+	va->va_end = addr + size;
+	if (va->va_end < addr || va->va_end < size || va->va_end > vend)
 		goto overflow;

 	va->va_start = addr;
-	va->va_end = addr + size;
 	va->flags = 0;
+
 	__insert_vmap_area(va);
 	free_vmap_cache = &va->rb_node;
 	spin_unlock(&vmap_area_lock);
--
1.8.1.2

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/