Mailing List Archive

[PATCH]: Use stop_machine_run in the Intel RNG driver
Replace call_smp_function with stop_machine_run in the Intel RNG driver.

CPU A has done read_lock(&lock)
CPU B has done write_lock_irq(&lock) and is waiting for A to release the lock.

A third CPU calls call_smp_function and issues the IPI. CPU A takes CPU C's
IPI. CPU B is waiting with interrupts disabled and does not see the IPI.
CPU C is stuck waiting for CPU B to respond to the IPI.

Deadlock.

The solution is to use stop_machine_run instead of call_smp_function
(call_smp_function should not be called in situations where the CPUs may
be suspended).

Signed-off-by: Prarit Bhargava <prarit@redhat.com>

--- linux-2.6.18.x86_64.orig/drivers/char/hw_random/intel-rng.c 2007-02-23 06:55:34.000000000 -0500
+++ linux-2.6.18.x86_64/drivers/char/hw_random/intel-rng.c 2007-02-26 04:59:36.000000000 -0500
@@ -24,10 +24,11 @@
* warranty of any kind, whether express or implied.
*/

-#include <linux/module.h>
+#include <linux/hw_random.h>
#include <linux/kernel.h>
+#include <linux/module.h>
#include <linux/pci.h>
-#include <linux/hw_random.h>
+#include <linux/stop_machine.h>
#include <asm/io.h>


@@ -217,30 +218,117 @@ static struct hwrng intel_rng = {
.data_read = intel_rng_data_read,
};

+struct intel_rng_hw {
+ struct pci_dev *dev;
+ void __iomem *mem;
+ u8 bios_cntl_off;
+ u8 bios_cntl_val;
+ u8 fwh_dec_en1_off;
+ u8 fwh_dec_en1_val;
+};
+
+static int __init intel_rng_hw_init(void *_intel_rng_hw)
+{
+ struct intel_rng_hw *intel_rng_hw = _intel_rng_hw;
+ u8 mfc, dvc;
+
+ /* interrupts disabled in stop_machine_run call */
+
+ if (!(intel_rng_hw->fwh_dec_en1_val & FWH_F8_EN_MASK))
+ pci_write_config_byte(intel_rng_hw->dev,
+ intel_rng_hw->fwh_dec_en1_off,
+ intel_rng_hw->fwh_dec_en1_val |
+ FWH_F8_EN_MASK);
+ if (!(intel_rng_hw->bios_cntl_val & BIOS_CNTL_WRITE_ENABLE_MASK))
+ pci_write_config_byte(intel_rng_hw->dev,
+ intel_rng_hw->bios_cntl_off,
+ intel_rng_hw->bios_cntl_val |
+ BIOS_CNTL_WRITE_ENABLE_MASK);
+
+ writeb(INTEL_FWH_RESET_CMD, intel_rng_hw->mem);
+ writeb(INTEL_FWH_READ_ID_CMD, intel_rng_hw->mem);
+ mfc = readb(intel_rng_hw->mem + INTEL_FWH_MANUFACTURER_CODE_ADDRESS);
+ dvc = readb(intel_rng_hw->mem + INTEL_FWH_DEVICE_CODE_ADDRESS);
+ writeb(INTEL_FWH_RESET_CMD, intel_rng_hw->mem);
+
+ if (!(intel_rng_hw->bios_cntl_val &
+ (BIOS_CNTL_LOCK_ENABLE_MASK|BIOS_CNTL_WRITE_ENABLE_MASK)))
+ pci_write_config_byte(intel_rng_hw->dev,
+ intel_rng_hw->bios_cntl_off,
+ intel_rng_hw->bios_cntl_val);
+ if (!(intel_rng_hw->fwh_dec_en1_val & FWH_F8_EN_MASK))
+ pci_write_config_byte(intel_rng_hw->dev,
+ intel_rng_hw->fwh_dec_en1_off,
+ intel_rng_hw->fwh_dec_en1_val);
+
+ if (mfc != INTEL_FWH_MANUFACTURER_CODE ||
+ (dvc != INTEL_FWH_DEVICE_CODE_8M &&
+ dvc != INTEL_FWH_DEVICE_CODE_4M)) {
+ printk(KERN_ERR PFX "FWH not detected\n");
+ return -ENODEV;
+ }

-#ifdef CONFIG_SMP
-static char __initdata waitflag;
+ return 0;
+}

-static void __init intel_init_wait(void *unused)
+static int __init intel_init_hw_struct(struct intel_rng_hw *intel_rng_hw,
+ struct pci_dev *dev)
{
- while (waitflag)
- cpu_relax();
+ intel_rng_hw->bios_cntl_val = 0xff;
+ intel_rng_hw->fwh_dec_en1_val = 0xff;
+ intel_rng_hw->dev = dev;
+
+ /* Check for Intel 82802 */
+ if (dev->device < 0x2640) {
+ intel_rng_hw->fwh_dec_en1_off = FWH_DEC_EN1_REG_OLD;
+ intel_rng_hw->bios_cntl_off = BIOS_CNTL_REG_OLD;
+ } else {
+ intel_rng_hw->fwh_dec_en1_off = FWH_DEC_EN1_REG_NEW;
+ intel_rng_hw->bios_cntl_off = BIOS_CNTL_REG_NEW;
+ }
+
+ pci_read_config_byte(dev, intel_rng_hw->fwh_dec_en1_off,
+ &intel_rng_hw->fwh_dec_en1_val);
+ pci_read_config_byte(dev, intel_rng_hw->bios_cntl_off,
+ &intel_rng_hw->bios_cntl_val);
+
+ if ((intel_rng_hw->bios_cntl_val &
+ (BIOS_CNTL_LOCK_ENABLE_MASK|BIOS_CNTL_WRITE_ENABLE_MASK))
+ == BIOS_CNTL_LOCK_ENABLE_MASK) {
+ static __initdata /*const*/ char warning[] =
+ KERN_WARNING PFX "Firmware space is locked read-only. "
+ KERN_WARNING PFX "If you can't or\n don't want to "
+ KERN_WARNING PFX "disable this in firmware setup, and "
+ KERN_WARNING PFX "if\n you are certain that your "
+ KERN_WARNING PFX "system has a functional\n RNG, try"
+ KERN_WARNING PFX "using the 'no_fwh_detect' option.\n";
+
+ if (no_fwh_detect)
+ return -ENODEV;
+ printk(warning);
+ return -EBUSY;
+ }
+
+ intel_rng_hw->mem = ioremap_nocache(INTEL_FWH_ADDR, INTEL_FWH_ADDR_LEN);
+ if (intel_rng_hw->mem == NULL)
+ return -EBUSY;
+
+ return 0;
}
-#endif
+

static int __init mod_init(void)
{
int err = -ENODEV;
- unsigned i;
+ int i;
struct pci_dev *dev = NULL;
- void __iomem *mem;
- unsigned long flags;
- u8 bios_cntl_off, fwh_dec_en1_off;
- u8 bios_cntl_val = 0xff, fwh_dec_en1_val = 0xff;
- u8 hw_status, mfc, dvc;
+ void __iomem *mem = mem;
+ u8 hw_status;
+ struct intel_rng_hw *intel_rng_hw;

for (i = 0; !dev && pci_tbl[i].vendor; ++i)
- dev = pci_get_device(pci_tbl[i].vendor, pci_tbl[i].device, NULL);
+ dev = pci_get_device(pci_tbl[i].vendor, pci_tbl[i].device,
+ NULL);

if (!dev)
goto out; /* Device not found. */
@@ -250,39 +338,18 @@ static int __init mod_init(void)
goto fwh_done;
}

- /* Check for Intel 82802 */
- if (dev->device < 0x2640) {
- fwh_dec_en1_off = FWH_DEC_EN1_REG_OLD;
- bios_cntl_off = BIOS_CNTL_REG_OLD;
- } else {
- fwh_dec_en1_off = FWH_DEC_EN1_REG_NEW;
- bios_cntl_off = BIOS_CNTL_REG_NEW;
- }
-
- pci_read_config_byte(dev, fwh_dec_en1_off, &fwh_dec_en1_val);
- pci_read_config_byte(dev, bios_cntl_off, &bios_cntl_val);
-
- if ((bios_cntl_val &
- (BIOS_CNTL_LOCK_ENABLE_MASK|BIOS_CNTL_WRITE_ENABLE_MASK))
- == BIOS_CNTL_LOCK_ENABLE_MASK) {
- static __initdata /*const*/ char warning[] =
- KERN_WARNING PFX "Firmware space is locked read-only. If you can't or\n"
- KERN_WARNING PFX "don't want to disable this in firmware setup, and if\n"
- KERN_WARNING PFX "you are certain that your system has a functional\n"
- KERN_WARNING PFX "RNG, try using the 'no_fwh_detect' option.\n";
-
+ intel_rng_hw = kmalloc(sizeof(*intel_rng_hw), GFP_KERNEL);
+ if (!intel_rng_hw) {
pci_dev_put(dev);
- if (no_fwh_detect)
- goto fwh_done;
- printk(warning);
- err = -EBUSY;
goto out;
}

- mem = ioremap_nocache(INTEL_FWH_ADDR, INTEL_FWH_ADDR_LEN);
- if (mem == NULL) {
+ err = intel_init_hw_struct(intel_rng_hw, dev);
+ if (err == -ENODEV) {
+ pci_dev_put(dev);
+ goto fwh_done;
+ } else if (err < 0) {
pci_dev_put(dev);
- err = -EBUSY;
goto out;
}

@@ -290,59 +357,17 @@ static int __init mod_init(void)
* Since the BIOS code/data is going to disappear from its normal
* location with the Read ID command, all activity on the system
* must be stopped until the state is back to normal.
+ *
+ * Use stop_machine_run because IPIs can be blocked by disabling
+ * interrupts.
*/
-#ifdef CONFIG_SMP
- set_mb(waitflag, 1);
- if (smp_call_function(intel_init_wait, NULL, 1, 0) != 0) {
- set_mb(waitflag, 0);
- pci_dev_put(dev);
- printk(KERN_ERR PFX "cannot run on all processors\n");
- err = -EAGAIN;
- goto err_unmap;
- }
-#endif
- local_irq_save(flags);
-
- if (!(fwh_dec_en1_val & FWH_F8_EN_MASK))
- pci_write_config_byte(dev,
- fwh_dec_en1_off,
- fwh_dec_en1_val | FWH_F8_EN_MASK);
- if (!(bios_cntl_val & BIOS_CNTL_WRITE_ENABLE_MASK))
- pci_write_config_byte(dev,
- bios_cntl_off,
- bios_cntl_val | BIOS_CNTL_WRITE_ENABLE_MASK);
-
- writeb(INTEL_FWH_RESET_CMD, mem);
- writeb(INTEL_FWH_READ_ID_CMD, mem);
- mfc = readb(mem + INTEL_FWH_MANUFACTURER_CODE_ADDRESS);
- dvc = readb(mem + INTEL_FWH_DEVICE_CODE_ADDRESS);
- writeb(INTEL_FWH_RESET_CMD, mem);
-
- if (!(bios_cntl_val &
- (BIOS_CNTL_LOCK_ENABLE_MASK|BIOS_CNTL_WRITE_ENABLE_MASK)))
- pci_write_config_byte(dev, bios_cntl_off, bios_cntl_val);
- if (!(fwh_dec_en1_val & FWH_F8_EN_MASK))
- pci_write_config_byte(dev, fwh_dec_en1_off, fwh_dec_en1_val);
-
- local_irq_restore(flags);
-#ifdef CONFIG_SMP
- /* Tell other CPUs to resume. */
- set_mb(waitflag, 0);
-#endif
-
- iounmap(mem);
+ err = stop_machine_run(intel_rng_hw_init, intel_rng_hw, NR_CPUS);
pci_dev_put(dev);
-
- if (mfc != INTEL_FWH_MANUFACTURER_CODE ||
- (dvc != INTEL_FWH_DEVICE_CODE_8M &&
- dvc != INTEL_FWH_DEVICE_CODE_4M)) {
- printk(KERN_ERR PFX "FWH not detected\n");
- err = -ENODEV;
- goto out;
- }
+ iounmap(intel_rng_hw->mem);
+ kfree(intel_rng_hw);
+ goto out;

fwh_done:
-
err = -ENOMEM;
mem = ioremap(INTEL_RNG_ADDR, INTEL_RNG_ADDR_LEN);
if (!mem)
@@ -352,22 +377,21 @@ fwh_done:
/* Check for Random Number Generator */
err = -ENODEV;
hw_status = hwstatus_get(mem);
- if ((hw_status & INTEL_RNG_PRESENT) == 0)
- goto err_unmap;
+ if ((hw_status & INTEL_RNG_PRESENT) == 0) {
+ iounmap(mem);
+ goto out;
+ }

printk(KERN_INFO "Intel 82802 RNG detected\n");
err = hwrng_register(&intel_rng);
if (err) {
printk(KERN_ERR PFX "RNG registering failed (%d)\n",
err);
- goto err_unmap;
+ iounmap(mem);
}
out:
return err;

-err_unmap:
- iounmap(mem);
- goto out;
}

static void __exit mod_exit(void)
--- linux-2.6.18.x86_64.orig/kernel/stop_machine.c 2006-09-19 23:42:06.000000000 -0400
+++ linux-2.6.18.x86_64/kernel/stop_machine.c 2007-02-23 07:16:29.000000000 -0500
@@ -1,8 +1,9 @@
-#include <linux/stop_machine.h>
-#include <linux/kthread.h>
-#include <linux/sched.h>
#include <linux/cpu.h>
#include <linux/err.h>
+#include <linux/kthread.h>
+#include <linux/module.h>
+#include <linux/sched.h>
+#include <linux/stop_machine.h>
#include <linux/syscalls.h>
#include <asm/atomic.h>
#include <asm/semaphore.h>
@@ -205,3 +206,4 @@ int stop_machine_run(int (*fn)(void *),

return ret;
}
+EXPORT_SYMBOL_GPL(stop_machine_run);
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Re: [PATCH]: Use stop_machine_run in the Intel RNG driver [ In reply to ]
Prarit Bhargava wrote:
> Replace call_smp_function with stop_machine_run in the Intel RNG driver.
>
> CPU A has done read_lock(&lock)
> CPU B has done write_lock_irq(&lock) and is waiting for A to release the lock.
>
> A third CPU calls call_smp_function and issues the IPI. CPU A takes CPU C's
> IPI. CPU B is waiting with interrupts disabled and does not see the IPI.
> CPU C is stuck waiting for CPU B to respond to the IPI.
>
> Deadlock.
>
> The solution is to use stop_machine_run instead of call_smp_function
> (call_smp_function should not be called in situations where the CPUs may
> be suspended).
>
>
>

Updated patch to latest-and-greatest git ...

P.
Re: [PATCH]: Use stop_machine_run in the Intel RNG driver [ In reply to ]
On Tue, 27 Feb 2007 07:22:00 -0500 Prarit Bhargava <prarit@redhat.com> wrote:

> Replace call_smp_function with stop_machine_run in the Intel RNG driver.
>
> CPU A has done read_lock(&lock)
> CPU B has done write_lock_irq(&lock) and is waiting for A to release the lock.
>
> A third CPU calls call_smp_function and issues the IPI. CPU A takes CPU C's
> IPI. CPU B is waiting with interrupts disabled and does not see the IPI.
> CPU C is stuck waiting for CPU B to respond to the IPI.
>
> Deadlock.

I think what you're describing here is just the standard old
smp_call_function() deadlock, rather than anything which is specific to
intel-rng, yes?

It is "well known" that you can't call smp_call_function() with local
interrupts disabled. In fact i386 will spit a warning if you try it.


intel-rng doesn't do that, but what it _does_ do is:

smp_call_function(..., wait = 0);
local_irq_disable();

so some CPUs will still be entering the IPI while this CPU has gone and
disabled interrupts, thus exposing us to the deadlock, yes?

In which case a suitable fix might be to make intel-rng spin until all the
other CPUs have entered intel_init_wait().

> The solution is to use stop_machine_run instead of call_smp_function
> (call_smp_function should not be called in situations where the CPUs may
> be suspended).

But that seems to be a nice change anyway. It took rather a lot of code
churn to do it, and it does find it necessary to export stop_machine_run()
to modules, but that seems OK too.

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Re: [PATCH]: Use stop_machine_run in the Intel RNG driver [ In reply to ]
> I think what you're describing here is just the standard old
> smp_call_function() deadlock, rather than anything which is specific to
> intel-rng, yes?
>
> It is "well known" that you can't call smp_call_function() with local
> interrupts disabled. In fact i386 will spit a warning if you try it.
>
>
> intel-rng doesn't do that, but what it _does_ do is:
>
> smp_call_function(..., wait = 0);
> local_irq_disable();
>
> so some CPUs will still be entering the IPI while this CPU has gone and
> disabled interrupts, thus exposing us to the deadlock, yes?
>

Not quite Andrew. This was a different and little more complicated.

The deadlock occurs because two CPUs are in contention over a rw_lock
and the "call_function" puts the CPUs in such a state that no forward
progress will be made until the calling CPU has completed it's code.

Here's a more detailed example (sorry for the cut-and-paste):

1. CPU A has done read_lock(&lock), and has acquired the lock.
2. CPU B has done write_lock_irq(&lock) and is waiting for A to release
the lock. CPU B has disabled interrupts while waiting for the interrupt:

void __lockfunc _write_lock_irq(rwlock_t *lock)
{
local_irq_disable();
preempt_disable();
rwlock_acquire(&lock->dep_map, 0, 0, _RET_IP_);
_raw_write_lock(lock);
}

3. CPU C issues smp_call_function, as in the case of the intel-rng driver:

set_mb(waitflag, 1);
smp_call_function(intel_init_wait, NULL, 1, 0);
...
// do some stuff with interrupts disabled
...
set_mb(waitflag, 0);

where

static char __initdata waitflag;

static void __init intel_init_wait(void *unused)
{
while (waitflag)
cpu_relax();
}

In this code the calling processor, C, has issued an IPI and disabled
interrupts on every processor except itself. When each processor takes
the IPI it runs intel_init_wait and waits in a tight loop until
waitflag is zero. ie) no forward progress on any CPU.

CPU C will not execute the code below the smp_call_function until all
processors have started (not completed!) the IPI function. From
call_smp_function:

cpus = num_online_cpus() - 1;
...

/* Send a message to all other CPUs and wait for them to respond */
send_IPI_allbutself(CALL_FUNCTION_VECTOR);

/* Wait for response */
while (atomic_read(&data.started) != cpus)
cpu_relax();

So CPU C is waiting here.

4. CPU A, which holds the lock sees the IPI and is in the
intel_init_wait code, happily waiting. CPU A has incremented
data.started. CPU A will stay in this loop until CPU C sets waitflag = 0.

5. CPU B, if you recall is _waiting with interrupts disabled_ for CPU A
to release the lock. It does not see the IPI because it has interrupts
disabled. It will not see the IPI until CPU A has released the lock.

6. CPU C is eventually only waiting for CPU B to do the final increment
of data.started = cpus. CPU B is waiting for CPU A to release the
lock. CPU A is executing a tight loop which it will not exit from until
CPU C can set waitflag to zero.

That's a 3-way deadlock.

So, the issue is placing the other CPUs in a state that they do not make
forward progress. The deadlock occurs before the calling CPU has
disabled interrupts in the code in step 3.

I also tested this code without the __init tags and explicitly coding
waitflag=0 to avoid the gcc only setting .bss section variables to zero
error that someone fixed last week. I also removed the code that
disabled interrupts on the calling processor which had no effect -- at
first I thought it was a simple interrupt issue ...

Maybe smp_call_function needs a written warning that the called function
should not "suspend" CPUs?

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