WSL2-Linux-Kernel/drivers/char
Suresh Siddha e49140120c crypto: padlock - fix VIA PadLock instruction usage with irq_ts_save/restore()
Wolfgang Walter reported this oops on his via C3 using padlock for
AES-encryption:

##################################################################

BUG: unable to handle kernel NULL pointer dereference at 000001f0
IP: [<c01028c5>] __switch_to+0x30/0x117
*pde = 00000000
Oops: 0002 [#1] PREEMPT
Modules linked in:

Pid: 2071, comm: sleep Not tainted (2.6.26 #11)
EIP: 0060:[<c01028c5>] EFLAGS: 00010002 CPU: 0
EIP is at __switch_to+0x30/0x117
EAX: 00000000 EBX: c0493300 ECX: dc48dd00 EDX: c0493300
ESI: dc48dd00 EDI: c0493530 EBP: c04cff8c ESP: c04cff7c
 DS: 007b ES: 007b FS: 0000 GS: 0033 SS: 0068
Process sleep (pid: 2071, ti=c04ce000 task=dc48dd00 task.ti=d2fe6000)
Stack: dc48df30 c0493300 00000000 00000000 d2fe7f44 c03b5b43 c04cffc8 00000046
       c0131856 0000005a dc472d3c c0493300 c0493470 d983ae00 00002696 00000000
       c0239f54 00000000 c04c4000 c04cffd8 c01025fe c04f3740 00049800 c04cffe0
Call Trace:
 [<c03b5b43>] ? schedule+0x285/0x2ff
 [<c0131856>] ? pm_qos_requirement+0x3c/0x53
 [<c0239f54>] ? acpi_processor_idle+0x0/0x434
 [<c01025fe>] ? cpu_idle+0x73/0x7f
 [<c03a4dcd>] ? rest_init+0x61/0x63
 =======================

Wolfgang also found out that adding kernel_fpu_begin() and kernel_fpu_end()
around the padlock instructions fix the oops.

Suresh wrote:

These padlock instructions though don't use/touch SSE registers, but it behaves
similar to other SSE instructions. For example, it might cause DNA faults
when cr0.ts is set. While this is a spurious DNA trap, it might cause
oops with the recent fpu code changes.

This is the code sequence  that is probably causing this problem:

a) new app is getting exec'd and it is somewhere in between
   start_thread() and flush_old_exec() in the load_xyz_binary()

b) At pont "a", task's fpu state (like TS_USEDFPU, used_math() etc) is
   cleared.

c) Now we get an interrupt/softirq which starts using these encrypt/decrypt
   routines in the network stack. This generates a math fault (as
   cr0.ts is '1') which sets TS_USEDFPU and restores the math that is
   in the task's xstate.

d) Return to exec code path, which does start_thread() which does
   free_thread_xstate() and sets xstate pointer to NULL while
   the TS_USEDFPU is still set.

e) At the next context switch from the new exec'd task to another task,
   we have a scenarios where TS_USEDFPU is set but xstate pointer is null.
   This can cause an oops during unlazy_fpu() in __switch_to()

Now:

1) This should happen with or with out pre-emption. Viro also encountered
   similar problem with out CONFIG_PREEMPT.

2) kernel_fpu_begin() and kernel_fpu_end() will fix this problem, because
   kernel_fpu_begin() will manually do a clts() and won't run in to the
   situation of setting TS_USEDFPU in step "c" above.

3) This was working before the fpu changes, because its a spurious
   math fault  which doesn't corrupt any fpu/sse registers and the task's
   math state was always in an allocated state.

With out the recent lazy fpu allocation changes, while we don't see oops,
there is a possible race still present in older kernels(for example,
while kernel is using kernel_fpu_begin() in some optimized clear/copy
page and an interrupt/softirq happens which uses these padlock
instructions generating DNA fault).

This is the failing scenario that existed even before the lazy fpu allocation
changes:

0. CPU's TS flag is set

1. kernel using FPU in some optimized copy  routine and while doing
kernel_fpu_begin() takes an interrupt just before doing clts()

2. Takes an interrupt and ipsec uses padlock instruction. And we
take a DNA fault as TS flag is still set.

3. We handle the DNA fault and set TS_USEDFPU and clear cr0.ts

4. We complete the padlock routine

5. Go back to step-1, which resumes clts() in kernel_fpu_begin(), finishes
the optimized copy routine and does kernel_fpu_end(). At this point,
we have cr0.ts again set to '1' but the task's TS_USEFPU is stilll
set and not cleared.

6. Now kernel resumes its user operation. And at the next context
switch, kernel sees it has do a FP save as TS_USEDFPU is still set
and then will do a unlazy_fpu() in __switch_to(). unlazy_fpu()
will take a DNA fault, as cr0.ts is '1' and now, because we are
in __switch_to(), math_state_restore() will get confused and will
restore the next task's FP state and will save it in prev tasks's FP state.
Remember, in __switch_to() we are already on the stack of the next task
but take a DNA fault for the prev task.

This causes the fpu leakage.

Fix the padlock instruction usage by calling them inside the
context of new routines irq_ts_save/restore(), which clear/restore cr0.ts
manually in the interrupt context. This will not generate spurious DNA
in the  context of the interrupt which will fix the oops encountered and
the possible FPU leakage issue.

Reported-and-bisected-by: Wolfgang Walter <wolfgang.walter@stwm.de>
Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
2008-08-13 22:02:26 +10:00
..
agp agp: fix SIS 5591/5592 wrong PCI id 2008-08-12 10:13:38 +10:00
hw_random crypto: padlock - fix VIA PadLock instruction usage with irq_ts_save/restore() 2008-08-13 22:02:26 +10:00
ip2 ip2: push BKL down for the firmware interface 2008-07-25 10:53:43 -07:00
ipmi ipmi/powerpc: Use linux/of_{device,platform}.h instead of asm 2008-07-30 15:26:53 +10:00
mwave mwave: ioctl BKL pushdown 2008-07-25 10:53:43 -07:00
pcmcia Merge branch 'for-jeff' of git://git.kernel.org/pub/scm/linux/kernel/git/chris/linux-2.6 into tmp 2008-08-07 04:05:46 -04:00
rio rio: push down the BKL into the firmware ioctl handler 2008-07-25 10:53:43 -07:00
tpm tpm: Use correct data types for sizes in tpm_write() and tpm_read() 2008-07-26 12:00:04 -07:00
xilinx_hwicap char/xilinx_hwicap/xilinx_hwicap.c: Removed duplicated include 2008-07-25 09:23:31 -07:00
.gitignore [MIPS] Remove IT8172-based platforms, ITE 8172G and Globespan IVR support. 2006-10-03 17:59:17 +01:00
ChangeLog
Kconfig powerpc/iseries: remove the old viocons driver 2008-08-07 18:07:10 -07:00
Makefile powerpc/iseries: remove the old viocons driver 2008-08-07 18:07:10 -07:00
amiserial.c m68k/amiserial: fix fallout of tty break handling rework 2008-08-06 13:24:41 -07:00
apm-emulation.c APM emulation: Notify about all suspend events, not just APM invoked ones (v2) 2008-07-16 23:27:02 +02:00
applicom.c char serial: switch drivers to ioremap_nocache 2008-04-30 08:29:48 -07:00
applicom.h
bfin-otp.c Blackfin char driver for Blackfin on-chip OTP memory (v3) 2008-04-25 08:04:56 +08:00
briq_panel.c briq_panel: BKL pushdown 2008-06-20 14:05:55 -06:00
bsr.c powerpc: Add driver for Barrier Synchronization Register 2008-07-15 12:24:55 +10:00
cd1865.h
consolemap.c Basic braille screen reader support 2008-04-30 08:29:52 -07:00
cp437.uni
cs5535_gpio.c Add a bunch of cycle_kernel_lock() calls 2008-06-20 14:05:53 -06:00
cyclades.c tty: rework break handling 2008-07-22 13:03:28 -07:00
defkeymap.c_shipped Fix default compose table initialization 2008-03-03 14:53:16 -08:00
defkeymap.map
digi1.h
digiFep1.h
digiPCI.h
ds1286.c ds1286: BKL pushdown 2008-06-20 14:05:56 -06:00
ds1302.c ds1302: push down the BKL into the driver ioctl code 2008-07-25 10:53:43 -07:00
ds1620.c [ARM] Move include/asm-arm/arch-* to arch/arm/*/include/mach 2008-08-07 09:55:48 +01:00
dsp56k.c dsp56k: Fix BKL pushdown 2008-07-26 13:22:56 -07:00
dtlk.c Add a bunch of cycle_kernel_lock() calls 2008-06-20 14:05:53 -06:00
efirtc.c drivers/char/efirtc.c: removed duplicated #include 2008-08-04 16:59:56 -07:00
epca.c Fix the epca driver to permit epca_setup() to be invoked from the kernel cmdline 2008-07-22 13:03:28 -07:00
epca.h epca: use tty_port 2008-07-20 17:12:36 -07:00
epcaconfig.h
esp.c tty: rework break handling 2008-07-22 13:03:28 -07:00
generic_nvram.c driver/char/generic_nvram: fix banner 2008-06-12 18:05:41 -07:00
generic_serial.c gs: use tty_port 2008-07-20 17:12:36 -07:00
genrtc.c genrtc: BKL pushdown 2008-06-20 14:05:57 -06:00
hangcheck-timer.c drivers/char/: Spelling fixes 2008-02-03 17:11:42 +02:00
hpet.c #if 0 hpet_unregister() 2008-07-25 10:53:43 -07:00
hvc_beat.c [POWERPC] hvcbeat: Fix buffer manipulation 2008-04-17 07:46:10 +10:00
hvc_console.c fix spinlock recursion in hvc_console 2008-08-12 17:52:55 +10:00
hvc_console.h powerpc: Move include files to arch/powerpc/include/asm 2008-08-04 12:02:00 +10:00
hvc_irq.c hvc_console: rework setup to replace irq functions with callbacks 2008-07-25 12:06:06 +10:00
hvc_iseries.c hvc_console: rework setup to replace irq functions with callbacks 2008-07-25 12:06:06 +10:00
hvc_rtas.c [POWERPC] hvc_rtas_init() must be __init 2008-02-14 22:11:02 +11:00
hvc_vio.c hvc_console: rework setup to replace irq functions with callbacks 2008-07-25 12:06:06 +10:00
hvc_xen.c hvc_console: rework setup to replace irq functions with callbacks 2008-07-25 12:06:06 +10:00
hvcs.c powerpc: Move include files to arch/powerpc/include/asm 2008-08-04 12:02:00 +10:00
hvsi.c drivers/char: replace remaining __FUNCTION__ occurrences 2008-04-30 08:29:54 -07:00
i8k.c i8k: make fan multiplier tunable with a module parameter 2008-05-01 08:04:00 -07:00
ip27-rtc.c ip27-rtc: BKL pushdown 2008-06-20 14:05:57 -06:00
isicom.c isicom: restore using hardware break support 2008-07-22 13:03:28 -07:00
istallion.c Merge git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/driver-core-2.6 2008-07-22 13:13:47 -07:00
keyboard.c Merge master.kernel.org:/pub/scm/linux/kernel/git/torvalds/linux-2.6 into next 2008-07-21 00:55:14 -04:00
lp.c device create: char: convert device_create to device_create_drvdata 2008-07-21 21:54:41 -07:00
mbcs.c mbcs: cdev lock_kernel() pushdown 2008-06-20 14:05:48 -06:00
mbcs.h MBCS: convert dmareadlock to mutex 2008-02-07 08:42:25 -08:00
mem.c use generic_access_phys for /dev/mem mappings 2008-07-24 10:47:15 -07:00
misc.c device create: char: convert device_create to device_create_drvdata 2008-07-21 21:54:41 -07:00
mmtimer.c mmtimer: Push BKL down into the ioctl handler 2008-07-17 11:34:49 -07:00
moxa.c tty: rework break handling 2008-07-22 13:03:28 -07:00
moxa.h Char: moxa, cleanup rx/tx 2008-04-30 08:29:43 -07:00
mspec.c mspec: convert nopfn to fault 2008-07-24 10:47:14 -07:00
mxser.c Char: mxser, ratelimit ioctl warning 2008-07-30 09:41:45 -07:00
mxser.h mxser: convert large macros to functions 2008-04-30 08:29:49 -07:00
n_hdlc.c n_hdlc: honor O_NONBLOCK on write 2008-07-22 13:03:28 -07:00
n_r3964.c tty: Ldisc revamp 2008-07-20 17:12:34 -07:00
n_tty.c tty: Ldisc revamp 2008-07-20 17:12:34 -07:00
nozomi.c drivers/char: replace remaining __FUNCTION__ occurrences 2008-04-30 08:29:54 -07:00
nsc_gpio.c [PATCH] struct path: convert char-drivers 2006-12-08 08:28:44 -08:00
nvram.c drivers/char/nvram.c: Removed duplicated include 2008-07-23 09:36:23 -07:00
nwbutton.c [PATCH] Char: timers cleanup 2007-02-12 09:48:30 -08:00
nwbutton.h IRQ: Maintain regs pointer globally rather than passing to IRQ handlers 2006-10-05 15:10:12 +01:00
nwflash.c [ARM] fix nwflash.c: 6ee8928d94 2008-07-26 16:29:24 +01:00
pc8736x_gpio.c Add a bunch of cycle_kernel_lock() calls 2008-06-20 14:05:53 -06:00
ppdev.c ppdev: wrap ioctl handler in driver and push lock down 2008-07-25 10:53:43 -07:00
ps3flash.c ps3: FLASH ROM Storage Driver 2007-07-21 17:49:16 -07:00
pty.c tty: Ldisc revamp 2008-07-20 17:12:34 -07:00
random.c PAGE_ALIGN(): correctly handle 64-bit values on 32-bit architectures 2008-07-24 10:47:21 -07:00
raw.c device create: char: convert device_create to device_create_drvdata 2008-07-21 21:54:41 -07:00
riscom8.c riscom8: Restore driver using new break functionality 2008-07-22 13:03:28 -07:00
riscom8.h tty: add more tty_port fields 2008-07-20 17:12:38 -07:00
riscom8_reg.h
rocket.c tty: rework break handling 2008-07-22 13:03:28 -07:00
rocket.h tty: add more tty_port fields 2008-07-20 17:12:38 -07:00
rocket_int.h tty: add more tty_port fields 2008-07-20 17:12:38 -07:00
rtc.c drivers/char/rtc.c: removed duplicated include 2008-08-12 16:07:30 -07:00
scc.h
scx200_gpio.c Add a bunch of cycle_kernel_lock() calls 2008-06-20 14:05:53 -06:00
selection.c tty: Ldisc revamp 2008-07-20 17:12:34 -07:00
ser_a2232.c m68k: gs: use tty_port fixes 2008-07-26 20:29:03 -07:00
ser_a2232.h
ser_a2232fw.ax
ser_a2232fw.h
serial167.c m68k: serial167 missing return value in cy_put_char() 2008-05-05 12:37:02 -07:00
snsc.c device create: char: convert device_create to device_create_drvdata 2008-07-21 21:54:41 -07:00
snsc.h Convert asm/semaphore.h users to linux/semaphore.h 2008-04-18 22:22:54 -04:00
snsc_event.c byteorder: don't directly include linux/byteorder/generic.h 2008-05-16 12:01:45 -07:00
sonypi.c sonypi: BKL pushdown 2008-07-02 15:06:25 -06:00
specialix.c specialix: restore driver using new break functionality 2008-07-22 13:03:28 -07:00
specialix_io8.h tty: add more tty_port fields 2008-07-20 17:12:38 -07:00
stallion.c stallion: removed unused variable 2008-07-24 10:47:30 -07:00
sx.c sx: push BKL down into the firmware ioctl handler 2008-07-25 10:53:43 -07:00
sx.h Char: char/serial, remove SERIAL_TYPE_NORMAL redefines 2008-02-07 08:42:33 -08:00
sxboards.h
sxwindow.h
synclink.c Merge branch 'for-jeff' of git://git.kernel.org/pub/scm/linux/kernel/git/chris/linux-2.6 into tmp 2008-08-07 04:05:46 -04:00
synclink_gt.c Merge branch 'for-jeff' of git://git.kernel.org/pub/scm/linux/kernel/git/chris/linux-2.6 into tmp 2008-08-07 04:05:46 -04:00
synclinkmp.c Merge branch 'for-jeff' of git://git.kernel.org/pub/scm/linux/kernel/git/chris/linux-2.6 into tmp 2008-08-07 04:05:46 -04:00
sysrq.c fix: "smp_call_function: get rid of the unused nonatomic/retry argument" 2008-06-27 11:52:45 +02:00
tb0219.c Add a bunch of cycle_kernel_lock() calls 2008-06-20 14:05:53 -06:00
tlclk.c tlckl: BKL pushdown 2008-06-20 14:05:51 -06:00
toshiba.c toshiba: use ioremap_cached 2008-04-30 23:15:34 +02:00
tty_audit.c [PATCH] split linux/file.h 2008-05-01 13:08:16 -04:00
tty_io.c Fix race/oops in tty layer after BKL pushdown 2008-08-11 10:37:34 -07:00
tty_ioctl.c tty: Ldisc revamp 2008-07-20 17:12:34 -07:00
tty_ldisc.c try harder to load tty ldisc driver 2008-08-01 12:50:15 -07:00
vc_screen.c device create: char: convert device_create to device_create_drvdata 2008-07-21 21:54:41 -07:00
viotape.c device create: char: convert device_create to device_create_drvdata 2008-07-21 21:54:41 -07:00
virtio_console.c virtio_console: use virtqueue notification for hvc_console 2008-07-25 12:06:06 +10:00
vme_scc.c m68k: gs: use tty_port fixes 2008-07-26 20:29:03 -07:00
vr41xx_giu.c Add a bunch of cycle_kernel_lock() calls 2008-06-20 14:05:53 -06:00
vt.c vt: Deadlock workaround 2008-08-04 17:12:07 -07:00
vt_ioctl.c vt_ioctl: Prepare for BKL push down 2008-04-30 08:29:40 -07:00