| Message ID | 1453344982-4529-1-git-send-email-stefan@agner.ch (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
Hi Thierry, This seems not to have made it into 4.6-rc1, any chance to get it still into 4.6? It seems also to resolve an issue when using cat on the debug file: # cat /sys/kernel/debug/regmap/40039000.pwm/registers [ 326.707319] Unhandled fault: external abort on non-linefetch (0x1008) at 0x9083e01c [ 326.715097] pgd = 8de50000 [ 326.717848] [9083e01c] *pgd=8f805811, *pte=40039653, *ppte=40039453 [ 326.724222] Internal error: : 1008 [#3] ARM [ 326.728445] Modules linked in: [ 326.731556] CPU: 0 PID: 19407 Comm: cat Tainted: G D 4.6.0-rc1 #359 This is gone when using REGCACHE_FLAT... -- Stefan On 2016-01-20 18:56, Stefan Agner wrote: > Use flat regmap cache to avoid lockdep warning at probe: > [ 0.697285] WARNING: CPU: 0 PID: 1 at kernel/locking/lockdep.c:2755 > lockdep_trace_alloc+0x15c/0x160() > [ 0.697449] DEBUG_LOCKS_WARN_ON(irqs_disabled_flags(flags)) > > The RB-tree regmap cache needs to allocate new space on first > writes. However, allocations in an atomic context (e.g. when a > spinlock is held) are not allowed. The function regmap_write > calls map->lock, which acquires a spinlock in the fast_io case. > Since the pwm-fsl-ftm driver uses MMIO, the regmap bus of type > regmap_mmio is being used which has fast_io set to true. > > The MMIO space of the pwm-fsl-ftm driver is reasonable condense, > hence using the much faster flat regmap cache is anyway the better > choice. > > Signed-off-by: Stefan Agner <stefan@agner.ch> > Cc: Mark Brown <broonie@kernel.org> > --- > drivers/pwm/pwm-fsl-ftm.c | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > > diff --git a/drivers/pwm/pwm-fsl-ftm.c b/drivers/pwm/pwm-fsl-ftm.c > index f9dfc8b..ed39e8e 100644 > --- a/drivers/pwm/pwm-fsl-ftm.c > +++ b/drivers/pwm/pwm-fsl-ftm.c > @@ -414,7 +414,7 @@ static const struct regmap_config fsl_pwm_regmap_config = { > > .max_register = FTM_PWMLOAD, > .volatile_reg = fsl_pwm_volatile_reg, > - .cache_type = REGCACHE_RBTREE, > + .cache_type = REGCACHE_FLAT, > }; > > static int fsl_pwm_probe(struct platform_device *pdev)
diff --git a/drivers/pwm/pwm-fsl-ftm.c b/drivers/pwm/pwm-fsl-ftm.c index f9dfc8b..ed39e8e 100644 --- a/drivers/pwm/pwm-fsl-ftm.c +++ b/drivers/pwm/pwm-fsl-ftm.c @@ -414,7 +414,7 @@ static const struct regmap_config fsl_pwm_regmap_config = { .max_register = FTM_PWMLOAD, .volatile_reg = fsl_pwm_volatile_reg, - .cache_type = REGCACHE_RBTREE, + .cache_type = REGCACHE_FLAT, }; static int fsl_pwm_probe(struct platform_device *pdev)
Use flat regmap cache to avoid lockdep warning at probe: [ 0.697285] WARNING: CPU: 0 PID: 1 at kernel/locking/lockdep.c:2755 lockdep_trace_alloc+0x15c/0x160() [ 0.697449] DEBUG_LOCKS_WARN_ON(irqs_disabled_flags(flags)) The RB-tree regmap cache needs to allocate new space on first writes. However, allocations in an atomic context (e.g. when a spinlock is held) are not allowed. The function regmap_write calls map->lock, which acquires a spinlock in the fast_io case. Since the pwm-fsl-ftm driver uses MMIO, the regmap bus of type regmap_mmio is being used which has fast_io set to true. The MMIO space of the pwm-fsl-ftm driver is reasonable condense, hence using the much faster flat regmap cache is anyway the better choice. Signed-off-by: Stefan Agner <stefan@agner.ch> Cc: Mark Brown <broonie@kernel.org> --- drivers/pwm/pwm-fsl-ftm.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)