linux-stable.git/drivers/base/regmap, branch v5.15.78

regmap-irq: Fix offset/index mismatch in read_sub_irq_data()

2022-06-29T07:03:25+00:00

[ Upstream commit 3f05010f243be06478a9b11cfce0ce994f5a0890 ]

We need to divide the sub-irq status register offset by register
stride to get an index for the status buffer to avoid an out of
bounds write when the register stride is greater than 1.

Fixes: a2d21848d921 ("regmap: regmap-irq: Add main status register support")
Signed-off-by: Aidan MacDonald 
Link: https://lore.kernel.org/r/20220620200644.1961936-3-aidanmacdonald.0x0@gmail.com
Signed-off-by: Mark Brown 
Signed-off-by: Sasha Levin

regmap-irq: Fix a bug in regmap_irq_enable() for type_in_mask chips

2022-06-29T07:03:25+00:00

[ Upstream commit 485037ae9a095491beb7f893c909a76cc4f9d1e7 ]

When enabling a type_in_mask irq, the type_buf contents must be
AND'd with the mask of the IRQ we're enabling to avoid enabling
other IRQs by accident, which can happen if several type_in_mask
irqs share a mask register.

Fixes: bc998a730367 ("regmap: irq: handle HW using separate rising/falling edge interrupts")
Signed-off-by: Aidan MacDonald 
Link: https://lore.kernel.org/r/20220620200644.1961936-2-aidanmacdonald.0x0@gmail.com
Signed-off-by: Mark Brown 
Signed-off-by: Sasha Levin

regmap-irq: Update interrupt clear register for proper reset

2022-03-02T10:48:01+00:00

[ Upstream commit d04ad245d67a3991dfea5e108e4c452c2ab39bac ]

With the existing logic where clear_ack is true (HW doesn’t support
auto clear for ICR), interrupt clear register reset is not handled
properly. Due to this only the first interrupts get processed properly
and further interrupts are blocked due to not resetting interrupt
clear register.

Example for issue case where Invert_ack is false and clear_ack is true:

    Say Default ISR=0x00 & ICR=0x00 and ISR is triggered with 2
    interrupts making ISR = 0x11.

    Step 1: Say ISR is set 0x11 (store status_buff = ISR). ISR needs to
            be cleared with the help of ICR once the Interrupt is processed.

    Step 2: Write ICR = 0x11 (status_buff), this will clear the ISR to 0x00.

    Step 3: Issue - In the existing code, ICR is written with ICR =
            ~(status_buff) i.e ICR = 0xEE -> This will block all the interrupts
            from raising except for interrupts 0 and 4. So expectation here is to
            reset ICR, which will unblock all the interrupts.

            if (chip->clear_ack) {
                 if (chip->ack_invert && !ret)
                  ........
                 else if (!ret)
                     ret = regmap_write(map, reg,
                            ~data->status_buf[i]);

So writing 0 and 0xff (when ack_invert is true) should have no effect, other
than clearing the ACKs just set.

Fixes: 3a6f0fb7b8eb ("regmap: irq: Add support to clear ack registers")
Signed-off-by: Prasad Kumpatla 
Reviewed-by: Charles Keepax 
Tested-by: Marek Szyprowski 
Link: https://lore.kernel.org/r/20220217085007.30218-1-quic_pkumpatl@quicinc.com
Signed-off-by: Mark Brown 
Signed-off-by: Sasha Levin

regmap: Call regmap_debugfs_exit() prior to _init()

2022-01-27T10:04:00+00:00

[ Upstream commit 530792efa6cb86f5612ff093333fec735793b582 ]

Since commit cffa4b2122f5 ("regmap: debugfs: Fix a memory leak when
calling regmap_attach_dev"), the following debugfs error is seen
on i.MX boards:

debugfs: Directory 'dummy-iomuxc-gpr@20e0000' with parent 'regmap' already present!

In the attempt to fix the memory leak, the above commit added a NULL check
for map->debugfs_name. For the first debufs entry, map->debugfs_name is NULL
and then the new name is allocated via kasprintf().

For the second debugfs entry, map->debugfs_name() is no longer NULL, so
it will keep using the old entry name and the duplicate name error is seen.

Quoting Mark Brown:

"That means that if the device gets freed we'll end up with the old debugfs
file hanging around pointing at nothing.
...
To be more explicit this means we need a call to regmap_debugfs_exit()
which will clean up all the existing debugfs stuff before we loose
references to it."

Call regmap_debugfs_exit() prior to regmap_debugfs_init() to fix
the problem.

Tested on i.MX6Q and i.MX6SX boards.

Fixes: cffa4b2122f5 ("regmap: debugfs: Fix a memory leak when calling regmap_attach_dev")
Suggested-by: Mark Brown 
Signed-off-by: Fabio Estevam 
Link: https://lore.kernel.org/r/20220107163307.335404-1-festevam@gmail.com
Signed-off-by: Mark Brown 
Signed-off-by: Sasha Levin

regmap: Fix possible double-free in regcache_rbtree_exit()

2021-10-12T10:48:43+00:00

In regcache_rbtree_insert_to_block(), when 'present' realloc failed,
the 'blk' which is supposed to assign to 'rbnode->block' will be freed,
so 'rbnode->block' points a freed memory, in the error handling path of
regcache_rbtree_init(), 'rbnode->block' will be freed again in
regcache_rbtree_exit(), KASAN will report double-free as follows:

BUG: KASAN: double-free or invalid-free in kfree+0xce/0x390
Call Trace:
 slab_free_freelist_hook+0x10d/0x240
 kfree+0xce/0x390
 regcache_rbtree_exit+0x15d/0x1a0
 regcache_rbtree_init+0x224/0x2c0
 regcache_init+0x88d/0x1310
 __regmap_init+0x3151/0x4a80
 __devm_regmap_init+0x7d/0x100
 madera_spi_probe+0x10f/0x333 [madera_spi]
 spi_probe+0x183/0x210
 really_probe+0x285/0xc30

To fix this, moving up the assignment of rbnode->block to immediately after
the reallocation has succeeded so that the data structure stays valid even
if the second reallocation fails.

Reported-by: Hulk Robot 
Fixes: 3f4ff561bc88b ("regmap: rbtree: Make cache_present bitmap per node")
Signed-off-by: Yang Yingliang 
Link: https://lore.kernel.org/r/20211012023735.1632786-1-yangyingliang@huawei.com
Signed-off-by: Mark Brown

Merge series "Use raw spinlocks in the ls-extirq driver" from Vladimir Oltean :

2021-08-26T12:40:35+00:00

The ls-extirq irqchip driver accesses regmap inside its implementation
of the struct irq_chip :: irq_set_type method, and currently regmap
only knows to lock using normal spinlocks. But the method above wants
raw spinlock context, so this isn't going to work and triggers a
"[ BUG: Invalid wait context ]" splat.

The best we can do given the arrangement of the code is to patch regmap
and the syscon driver: regmap to support raw spinlocks, and syscon to
request them on behalf of its ls-extirq consumer.

Link: https://lore.kernel.org/lkml/20210825135438.ubcuxm5vctt6ne2q@skbuf/T/#u

Vladimir Oltean (2):
  regmap: teach regmap to use raw spinlocks if requested in the config
  mfd: syscon: request a regmap with raw spinlocks for some devices

 drivers/base/regmap/internal.h |  4 ++++
 drivers/base/regmap/regmap.c   | 35 +++++++++++++++++++++++++++++-----
 drivers/mfd/syscon.c           | 16 ++++++++++++++++
 include/linux/regmap.h         |  2 ++
 4 files changed, 52 insertions(+), 5 deletions(-)

--
2.25.1

base-commit: 6efb943b8616ec53a5e444193dccf1af9ad627b5

regmap: teach regmap to use raw spinlocks if requested in the config

2021-08-26T11:07:32+00:00

Some drivers might access regmap in a context where a raw spinlock is
held. An example is drivers/irqchip/irq-ls-extirq.c, which calls
regmap_update_bits() from struct irq_chip :: irq_set_type, which is a
method called by __irq_set_trigger() under the desc->lock raw spin lock.

Since desc->lock is a raw spin lock and the regmap internal lock for
mmio is a plain spinlock (which can become sleepable on RT), this is an
invalid locking scheme and we get a splat stating that this is a
"[ BUG: Invalid wait context ]".

It seems reasonable for regmap to have an option use a raw spinlock too,
so add that in the config such that drivers can request it.

Suggested-by: Mark Brown 
Signed-off-by: Vladimir Oltean 
Link: https://lore.kernel.org/r/20210825205041.927788-2-vladimir.oltean@nxp.com
Signed-off-by: Mark Brown

regmap: allow const array for {devm_,}regmap_field_bulk_alloc reg_fields

2021-08-02T11:21:32+00:00

The reg_fields array fed to {devm_}regmap_field_bulk_alloc is currently
not const, which is not correct on semantics (the functions shouldn't
change reg_field contents) and prevents pre-defined const reg_field
array to be used.

As the implementation of this function doesn't change the content of it,
just add const to its prototype.

Signed-off-by: Icenowy Zheng 
Link: https://lore.kernel.org/r/20210802063741.76301-1-icenowy@sipeed.com
Signed-off-by: Mark Brown

regmap: Prefer unsigned int to bare use of unsigned

2021-07-11T22:48:43+00:00

Fix checkpatch warnings:
    WARNING: Prefer 'unsigned int' to bare use of 'unsigned'

Signed-off-by: Jinchao Wang 
Link: https://lore.kernel.org/r/20210628171907.63646-1-wjc@cdjrlc.com
Signed-off-by: Mark Brown

regmap: fix the offset of register error log

2021-07-11T22:48:42+00:00

This patch fixes the offset of register error log
by using regmap_get_offset().

Signed-off-by: Jeongtae Park 
Link: https://lore.kernel.org/r/20210701142630.44936-1-jeongtae.park@gmail.com
Signed-off-by: Mark Brown