Re: next-20170515: WARNING: CPU: 0 PID: 1 at arch/x86/mm/dump_pagetables.c:236 note_page+0x630/0x7e0

["Luis R. Rodriguez" <mcgrof@kernel.org>]

On Fri, May 19, 2017 at 05:40:16PM +0200, Luis R. Rodriguez wrote:
> On Fri, May 19, 2017 at 05:08:02AM +0200, Luis R. Rodriguez wrote:
> > On Fri, May 19, 2017 at 02:44:14AM +0200, Luis R. Rodriguez wrote:
> > > On Wed, May 17, 2017 at 10:53:06AM -0700, Kees Cook wrote:
> > > > On Wed, May 17, 2017 at 9:40 AM, Luis R. Rodriguez <mcgrof@kernel.org> wrote:
> > > > > Yes, but I had killed that boot session again, so upon my next boot
> > > > > I had a different layout, the ASLR gap was much larger:
> > > > >
> > > > > ---[ Modules ]---
> > > > > 0xffffffffc0000000-0xffffffffc01b0000        1728K                               pte
> > > > > 0xffffffffc01b0000-0xffffffffc01b1000           4K     RW                 GLB x  pte
> > > > > 0xffffffffc01b1000-0xffffffffc01b2000           4K                               pte
> > > > > 0xffffffffc01b2000-0xffffffffc01c6000          80K     ro                 GLB x  pte
> > > > > 0xffffffffc01c6000-0xffffffffc01cc000          24K     ro                 GLB NX pte
> > > > > 0xffffffffc01cc000-0xffffffffc01d5000          36K     RW                 GLB NX pte
> > > > >
> > > > > As you can guess if we follow similar pattern the RW hole is the one this boot
> > > > > warned about:
> > > > >
> > > > > [    1.450483] x86/mm: Found insecure W+X mapping at address ffffffffc01b0000/0xffffffffc01b0000
> > > > > [    1.451280] ------------[ cut here ]------------
> > > > > [    1.451721] WARNING: CPU: 1 PID: 1 at arch/x86/mm/dump_pagetables.c:236 note_page+0x630/0x7e0
> > > > > [    1.452499] Modules linked in:
> > > > > [    1.452791] CPU: 1 PID: 1 Comm: swapper/0 Not tainted 4.12.0-rc1-next-20170515+ #145
> > > > >
> > > > > I checked and indeed 0xffffffffc01b2000 is part of a module, it was not the first one
> > > > > on the /proc/modules list but then again /proc/modules does not seem to have a specific
> > > > > order other than perhaps being pegged into a linked list of modules once they go live,
> > > > > and it seems its typically output backwards from when that happened, sorting that
> > > > > by address we get:
> > > > 
> > > > Right, sorry, I'd expect it at the bottom of the list in
> > > > /proc/modules, but that's fine, it's there.
> > > > 
> > > > >
> > > > > root@piggy:~# cat /proc/modules | sort -k 6 | head -3
> > > > > e1000 143360 0 - Live 0xffffffffc01b2000 (E)
> > > > > mbcache 16384 1 ext4, Live 0xffffffffc01d6000 (E)
> > > > > scsi_mod 217088 4 sg,sr_mod,sd_mod,libata, Live 0xffffffffc01df000 (E)
> > > > >
> > > > > And this then seems to be the first module loaded:
> > > > >
> > > > > e1000 143360 0 - Live 0xffffffffc01b2000 (E)
> > > > >
> > > > > The output of dmesg seems to confirm this as per the list of modules sorted
> > > > > as per above.
> > > > >
> > > > >> Something touched the module gap and left is RW+x...
> > > > >
> > > > > Lemme try booting with e1000 renamed to e1000.ko.ignore and see how that goes.
> > > > 
> > > > Is it possible a module got loaded before e1000 and then unloaded?
> > > > That seems odd, but maybe unload isn't cleaning up?
> > > > 
> > > > >> Are you able to bisect this?
> > > > >
> > > > > This issue has been present for a while so since I recall this I might be
> > > > > able to reduce the number of needed target kernels to bisect. Lemme tinker
> > > > > a bit and if no clear culprit comes up then will try bisect.
> > > > 
> > > > Okay, thanks!
> > > 
> > > Sorry to report that this issue is present since the feature's addition. So
> > > the issue is there since its addition and is still present today. *But* it
> > > may also be a configuration issue, given I have booted this guest *without*
> > > this issue ...
> > > 
> > > So:
> > > 
> > > git checkout -b WX e1a58320a38dfa72be48a0f1a3a92273663ba6db
> > > 
> > > That boots with the warning. To help debug further I've minimized my modules
> > > to only a few: scsi_mod, e1000, libata.
> > > 
> > > I suspect at this point this is not the fault of a particular module but
> > > instead just an accounting semantic (>= or <= on an edge) but let's see.
> > > 
> > > I now boot on 4.3.0-rc3 on commit (e1a58320a38df ("x86/mm: Warn on W^X
> > > mappings") and I with:
> > > 
> > > [    0.949435] ------------[ cut here ]------------                             
> > > [    0.949992] WARNING: CPU: 2 PID: 1 at arch/x86/mm/dump_pagetables.c:225 note_page+0x635/0x7e0()
> > > [    0.950996] x86/mm: Found insecure W+X mapping at address ffffffffc0000000/0xffffffffc0000000
> > > [    0.951814] Modules linked in:                                               
> > > [    0.952123] CPU: 2 PID: 1 Comm: swapper/0 Not tainted 4.3.0-rc3-FINAL-TEST-WITH-WX-NOFLOPPY+ #365
> > > [    0.952929] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.10.2-0-g5f4c7b1-prebuilt.qemu-project.org 04/01/2014
> > > [    0.954033]  0000000000000000 000000001f722925 ffff88013a5d7d40 ffffffff812ff335
> > > [    0.954742]  ffff88013a5d7d88 ffff88013a5d7d78 ffffffff81079be2 ffff88013a5d7e90
> > > [    0.955522]  0000000000000000 0000000000000004 0000000000000000 0000000000000000
> > > [    0.956256] Call Trace:                                                      
> > > [    0.956496]  [<ffffffff812ff335>] dump_stack+0x44/0x5f                       
> > > [    0.956953]  [<ffffffff81079be2>] warn_slowpath_common+0x82/0xc0             
> > > [    0.957519]  [<ffffffff81079c7c>] warn_slowpath_fmt+0x5c/0x80                
> > > [    0.958066]  [<ffffffff8106c155>] note_page+0x635/0x7e0                      
> > > [    0.958595]  [<ffffffff8106c5eb>] ptdump_walk_pgd_level_core+0x2eb/0x410     
> > > [    0.959219]  [<ffffffff8106c7b7>] ptdump_walk_pgd_level_checkwx+0x17/0x20    
> > > [    0.959856]  [<ffffffff8106260d>] mark_rodata_ro+0xed/0x100                  
> > > [    0.960372]  [<ffffffff815aa7d0>] ? rest_init+0x80/0x80                      
> > > [    0.960869]  [<ffffffff815aa7ed>] kernel_init+0x1d/0xe0                      
> > > [    0.961358]  [<ffffffff815b798f>] ret_from_fork+0x3f/0x70                    
> > > [    0.961900]  [<ffffffff815aa7d0>] ? rest_init+0x80/0x80                      
> > > [    0.962389] ---[ end trace 6125ebcb24c9e3d0 ]---                             
> > > [    0.962822] x86/mm: Checked W+X mappings: FAILED, 1 W+X pages found.         
> > >                                                                                 
> > >                                                                                 
> > > ---[ High Kernel Mapping ]---                                                   
> > > 0xffffffff80000000-0xffffffff81000000          16M                               pmd
> > > 0xffffffff81000000-0xffffffff81600000           6M     ro         PSE     GLB x  pmd
> > > 0xffffffff81600000-0xffffffff81a00000           4M     ro         PSE     GLB NX pmd
> > > 0xffffffff81a00000-0xffffffff81c00000           2M     RW                 GLB NX pte
> > > 0xffffffff81c00000-0xffffffff82200000           6M     RW         PSE     GLB NX pmd
> > > 0xffffffff82200000-0xffffffff82400000           2M     RW                 GLB NX pte
> > > 0xffffffff82400000-0xffffffffc0000000         988M                               pmd
> > > ---[ Modules ]---                                                               
> > > 0xffffffffc0000000-0xffffffffc0001000           4K     RW                 GLB x  pte
> > > 0xffffffffc0001000-0xffffffffc0002000           4K                               pte
> > > 0xffffffffc0002000-0xffffffffc0039000         220K     RW                 GLB x  pte
> > > 
> > > root@piggy:~# cat /proc/modules | sort -k 6 | head -3                           
> > > scsi_mod 221979 4 sg,sd_mod,sr_mod,libata, Live 0xffffffffc0002000 (E)          
> > > e1000 127757 0 - Live 0xffffffffc004d000 (E)                                    
> > > libata 229931 2 ata_generic,ata_piix, Live 0xffffffffc0076000 (E) 
> > > 
> > > So that 4K RW seems suspect of getting used for allocation purpose on edge
> > > for a particular reason and it also happens to be on the edge of the high
> > > kernel mapping. Could it be the boundary semantic issue ?
> > > 
> > > For instance can it be that since 0xffffffffc0002000 is given to the first
> > > module by the allocator, scsi_mod, and since that address is *technically*
> > > part of two boundaries we get a splat ?
> > > 
> > > 0xffffffffc0001000-0xffffffffc0002000           4K                               pte
> > > 0xffffffffc0002000-0xffffffffc0039000         220K     RW                 GLB x  pte
> > 
> > Note on the latest linux-next and on the commit that introduced this the config
> > and kernel yields only *one* page:
> > 
> > x86/mm: Checked W+X mappings: FAILED, 1 W+X pages found.
> > 
> > I believe this is more indications my suspicion might be right.
> 
> If the following is a legit forced way to get query the kernel to ask it 
> who owns a page then perhaps this technique can be used in the future to
> figure out who the hell caused this. Catalin, can you confirm? In this
> case this is perhaps not a leaked page but I am trying to abuse the
> kmemleak debugfs API to query who allocated the page. Is that fine?
> 
> [    0.916771] WARNING: CPU: 0 PID: 1 at arch/x86/mm/dump_pagetables.c:235 note_page+0x63c/0x7e0
> [    0.917636] x86/mm: Found insecure W+X mapping at address ffffffffc03d5000/0xffffffffc03d5000
> [    0.918502] Modules linked in:
> [    0.918819] CPU: 0 PID: 1 Comm: swapper/0 Not tainted 4.11.0-mcgrof-force-config #340
> [    0.919631] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.10.2-0-g5f4c7b1-prebuilt.qemu-project.org 04/01/2014
> [    0.920011] Call Trace:
> [    0.920011]  dump_stack+0x63/0x81
> [    0.920011]  __warn+0xcb/0xf0
> [    0.920011]  warn_slowpath_fmt+0x5a/0x80
> [    0.920011]  note_page+0x63c/0x7e0
> [    0.920011]  ptdump_walk_pgd_level_core+0x3b1/0x460
> [    0.920011]  ? 0xffffffff86c00000
> [    0.920011]  ptdump_walk_pgd_level_checkwx+0x17/0x20
> [    0.920011]  mark_rodata_ro+0xf4/0x100
> [    0.920011]  ? rest_init+0x80/0x80
> [    0.920011]  kernel_init+0x2a/0x100
> [    0.920011]  ret_from_fork+0x2c/0x40
> [    0.925474] ---[ end trace dca00cd779490a2b ]---
> [    0.925959] x86/mm: Checked W+X mappings: FAILED, 1 W+X pages found.
> 
> echo dump=0xffffffffc03d5000 > /sys/kernel/debug/kmemleak
> dmesg | tail
> 
> [   49.209565] kmemleak: Object 0xffffffffc03d5000 (size 335):
> [   49.210814] kmemleak:   comm "swapper/0", pid 1, jiffies 4294892440
> [   49.212148] kmemleak:   min_count = 2
> [   49.212852] kmemleak:   count = 0
> [   49.213363] kmemleak:   flags = 0x1
> [   49.213363] kmemleak:   checksum = 0
> [   49.213363] kmemleak:   backtrace:
> [   49.213363]      kmemleak_alloc+0x4a/0xa0
> [   49.213363]      __vmalloc_node_range+0x20a/0x2b0
> [   49.213363]      module_alloc+0x67/0xc0
> [   49.213363]      arch_ftrace_update_trampoline+0xba/0x260
> [   49.213363]      ftrace_startup+0x90/0x210
> [   49.213363]      register_ftrace_function+0x4b/0x60
> [   49.213363]      arm_kprobe+0x84/0xe0
> [   49.213363]      register_kprobe+0x56e/0x5b0
> [   49.213363]      init_test_probes+0x61/0x560
> [   49.213363]      init_kprobes+0x1e3/0x206
> [   49.213363]      do_one_initcall+0x52/0x1a0
> [   49.213363]      kernel_init_freeable+0x178/0x200
> [   49.213363]      kernel_init+0xe/0x100
> [   49.213363]      ret_from_fork+0x2c/0x40
> [   49.213363]      0xffffffffffffffff

Aha! And the winner is:

CONFIG_KPROBES_SANITY_TEST

I confirm disabling it on 4.3.0-rc3 and on linux-next next-20170519 avoids the WARN.
I also can confirm using the 'echo dump=mem-area > /sys/kernel/debug/kmemleak' yields
the same trace for both of these kernels.

So -- the above kmemleak hack seems to actually work to seek who owns that page.

Now to figure out how the hell kernel/test_kprobes.c screws around with things.

  Luis