Flushing system-wide workqueues is dangerous and will be forbidden.
Replace system_wq with local atmel_wq.
Link: https://lkml.kernel.org/r/[email protected]
Signed-off-by: Tetsuo Handa <[email protected]>
---
Note: This patch is only compile tested.
drivers/crypto/atmel-ecc.c | 2 +-
drivers/crypto/atmel-i2c.c | 24 +++++++++++++++++++++++-
drivers/crypto/atmel-i2c.h | 1 +
drivers/crypto/atmel-sha204a.c | 2 +-
4 files changed, 26 insertions(+), 3 deletions(-)
diff --git a/drivers/crypto/atmel-ecc.c b/drivers/crypto/atmel-ecc.c
index 333fbefbbccb..59a57279e77b 100644
--- a/drivers/crypto/atmel-ecc.c
+++ b/drivers/crypto/atmel-ecc.c
@@ -398,7 +398,7 @@ static int __init atmel_ecc_init(void)
static void __exit atmel_ecc_exit(void)
{
- flush_scheduled_work();
+ atmel_i2c_flush_queue();
i2c_del_driver(&atmel_ecc_driver);
}
diff --git a/drivers/crypto/atmel-i2c.c b/drivers/crypto/atmel-i2c.c
index 6fd3e969211d..226c55bfb9d6 100644
--- a/drivers/crypto/atmel-i2c.c
+++ b/drivers/crypto/atmel-i2c.c
@@ -263,6 +263,8 @@ static void atmel_i2c_work_handler(struct work_struct *work)
work_data->cbk(work_data, work_data->areq, status);
}
+static struct workqueue_struct *atmel_wq;
+
void atmel_i2c_enqueue(struct atmel_i2c_work_data *work_data,
void (*cbk)(struct atmel_i2c_work_data *work_data,
void *areq, int status),
@@ -272,10 +274,16 @@ void atmel_i2c_enqueue(struct atmel_i2c_work_data *work_data,
work_data->areq = areq;
INIT_WORK(&work_data->work, atmel_i2c_work_handler);
- schedule_work(&work_data->work);
+ queue_work(atmel_wq, &work_data->work);
}
EXPORT_SYMBOL(atmel_i2c_enqueue);
+void atmel_i2c_flush_queue(void)
+{
+ flush_workqueue(atmel_wq);
+}
+EXPORT_SYMBOL(atmel_i2c_flush_queue);
+
static inline size_t atmel_i2c_wake_token_sz(u32 bus_clk_rate)
{
u32 no_of_bits = DIV_ROUND_UP(TWLO_USEC * bus_clk_rate, USEC_PER_SEC);
@@ -372,6 +380,20 @@ int atmel_i2c_probe(struct i2c_client *client, const struct i2c_device_id *id)
}
EXPORT_SYMBOL(atmel_i2c_probe);
+static int __init atmel_i2c_init(void)
+{
+ atmel_wq = alloc_workqueue("atmel_wq", 0, 0);
+ return atmel_wq ? 0 : -ENOMEM;
+}
+
+static void __exit atmel_i2c_exit(void)
+{
+ destroy_workqueue(atmel_wq);
+}
+
+module_init(atmel_i2c_init);
+module_exit(atmel_i2c_exit);
+
MODULE_AUTHOR("Tudor Ambarus <[email protected]>");
MODULE_DESCRIPTION("Microchip / Atmel ECC (I2C) driver");
MODULE_LICENSE("GPL v2");
diff --git a/drivers/crypto/atmel-i2c.h b/drivers/crypto/atmel-i2c.h
index 63b97b104f16..48929efe2a5b 100644
--- a/drivers/crypto/atmel-i2c.h
+++ b/drivers/crypto/atmel-i2c.h
@@ -173,6 +173,7 @@ void atmel_i2c_enqueue(struct atmel_i2c_work_data *work_data,
void (*cbk)(struct atmel_i2c_work_data *work_data,
void *areq, int status),
void *areq);
+void atmel_i2c_flush_queue(void);
int atmel_i2c_send_receive(struct i2c_client *client, struct atmel_i2c_cmd *cmd);
diff --git a/drivers/crypto/atmel-sha204a.c b/drivers/crypto/atmel-sha204a.c
index c96c14e7dab1..2168f877bd43 100644
--- a/drivers/crypto/atmel-sha204a.c
+++ b/drivers/crypto/atmel-sha204a.c
@@ -159,7 +159,7 @@ static int __init atmel_sha204a_init(void)
static void __exit atmel_sha204a_exit(void)
{
- flush_scheduled_work();
+ atmel_i2c_flush_queue();
i2c_del_driver(&atmel_sha204a_driver);
}
--
2.32.0
On Wed, Apr 20, 2022 at 10:27:11AM +0900, Tetsuo Handa wrote:
>
> +module_init(atmel_i2c_init);
> +module_exit(atmel_i2c_exit);
What if everything is built-in? Shouldn't this be moved to a place
earlier than module_init?
Thanks,
--
Email: Herbert Xu <[email protected]>
Home Page: http://gondor.apana.org.au/~herbert/
PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt
On 2022/04/29 14:30, Herbert Xu wrote:
> On Wed, Apr 20, 2022 at 10:27:11AM +0900, Tetsuo Handa wrote:
>>
>> +module_init(atmel_i2c_init);
>> +module_exit(atmel_i2c_exit);
>
> What if everything is built-in? Shouldn't this be moved to a place
> earlier than module_init?
Since drivers/crypto/Makefile has lines in
obj-$(CONFIG_CRYPTO_DEV_ATMEL_I2C) += atmel-i2c.o
obj-$(CONFIG_CRYPTO_DEV_ATMEL_ECC) += atmel-ecc.o
obj-$(CONFIG_CRYPTO_DEV_ATMEL_SHA204A) += atmel-sha204a.o
order (which will be used as link order for built-in.o), module_init()
is processed in this order. Also, module_exit() is no-op if built-in.
Therefore, I think there is no need to explicitly boost the priority
of atmel_i2c_init().
On Fri, Apr 29, 2022 at 03:26:07PM +0900, Tetsuo Handa wrote:
>
> Since drivers/crypto/Makefile has lines in
>
> obj-$(CONFIG_CRYPTO_DEV_ATMEL_I2C) += atmel-i2c.o
> obj-$(CONFIG_CRYPTO_DEV_ATMEL_ECC) += atmel-ecc.o
> obj-$(CONFIG_CRYPTO_DEV_ATMEL_SHA204A) += atmel-sha204a.o
>
> order (which will be used as link order for built-in.o), module_init()
> is processed in this order. Also, module_exit() is no-op if built-in.
> Therefore, I think there is no need to explicitly boost the priority
> of atmel_i2c_init().
If we're going to rely on the Makefile ordering please make it
explicit and add a comment in the Makefile. Otherwise this is way
too fragile.
Thanks,
--
Email: Herbert Xu <[email protected]>
Home Page: http://gondor.apana.org.au/~herbert/
PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt