| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Date: Thu, 16 Mar 2023 15:38:03 +0100
From: "Fabio M. De Francesco" <fmdefrancesco@...il.com>
To: outreachy@...ts.linux.dev,
Khadija Kamran <kamrankhadijadj@...il.com>
Cc: Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
linux-staging@...ts.linux.dev, linux-kernel@...r.kernel.org
Subject: Re: [PATCH v5] staging: axis-fifo: initialize timeouts in probe only
Khadija,
Just saw your v5 patch and Greg's two replies.
For v6 you will need to change the subject to "[PATCH v6] staging: axis-fifo:
initialize timeouts in init only" to indicate that you are doing assignments
in axis_fifo_init().
Don't forget to extend the version log with "Changes in v6:" and clarify that
v5 had a different "Object" (you should probably also add a link to the v5
patch in lore: https://lore.kernel.org/lkml /ZBMR4s8xyHGqMm72@...dija-virtual-
machine/). When the "Subject" changes, readers may not find the previous
versions easily.
On giovedì 16 marzo 2023 13:56:02 CET Khadija Kamran wrote:
> Module parameter, read_timeout, can only be set at the loading time. As
> it can only be modified once, initialize read_timeout once in the probe
Substitute "probe" with "init".
> function.
>
> As a result, only use read_timeout as the last argument in
> wait_event_interruptible_timeout() call.
This two sentences are not much clear. I'd merge and rework:
"Initialize the module parameters read_timeout and write_timeout once in
init().
Module parameters can only be set once and cannot be modified later, so we
don't need to evaluate them again when passing the parameters to
wait_event_interruptible_timeout()."
>
> Convert datatpe
s/datatpe/type/
> of read_timeout
of {read,write}_timeout
> from 'int' to 'long int' because
> implicit conversion of 'long int' to 'int' in statement 'read_timeout =
> MAX_SCHEDULE_TIMEOUT' results in an overflow warning.
We don't care too much about the warning themselves: I mean, it overflows and
you must avoid it to happen (as you are doing with the changes of types), not
merely be interested in avoiding the warning. "[] results in an overflow." is
all we care about.
Add also the previous paragraph in the last part of the commit message.
> Perform same steps formodule parameter, write_timeout.
And instead delete the this last phrase.
> Suggested-by: Greg Kroah-Hartman <gregkh@...uxfoundation.org>
> Signed-off-by: Khadija Kamran <kamrankhadijadj@...il.com>
> ---
>
> Changes in v5:
> - Convert timeout's datatype from int to long.
> Changes in v4:
> - Initialize timeouts once as suggested by Greg; this automatically
> fixes the indentation problems.
> - Change the subject and description.
> Changes in v3:
> - Fix grammatical mistakes
> - Do not change the second argument's indentation in split lines
> Changes in v2:
> - Instead of matching alignment to open parenthesis, align second and
> the last argument instead.
> - Change the subject to 'remove tabs to align arguments'.
> - Use imperative language in subject and description
>
> drivers/staging/axis-fifo/axis-fifo.c | 26 ++++++++++++++++----------
> 1 file changed, 16 insertions(+), 10 deletions(-)
>
> diff --git a/drivers/staging/axis-fifo/axis-fifo.c
> b/drivers/staging/axis-fifo/axis-fifo.c index dfd2b357f484..d667dc80df47
> 100644
> --- a/drivers/staging/axis-fifo/axis-fifo.c
> +++ b/drivers/staging/axis-fifo/axis-fifo.c
> @@ -103,17 +103,17 @@
> * globals
> * ----------------------------
> */
> -static int read_timeout = 1000; /* ms to wait before read() times out */
> -static int write_timeout = 1000; /* ms to wait before write() times out */
> +static long read_timeout = 1000; /* ms to wait before read() times out */
> +static long write_timeout = 1000; /* ms to wait before write() times out */
>
> /* ----------------------------
> * module command-line arguments
> * ----------------------------
> */
>
> -module_param(read_timeout, int, 0444);
> +module_param(read_timeout, long, 0444);
> MODULE_PARM_DESC(read_timeout, "ms to wait before blocking read() timing
out;
> set to -1 for no timeout"); -module_param(write_timeout, int, 0444);
> +module_param(write_timeout, long, 0444);
> MODULE_PARM_DESC(write_timeout, "ms to wait before blocking write() timing
> out; set to -1 for no timeout");
>
> /* ----------------------------
> @@ -384,9 +384,7 @@ static ssize_t axis_fifo_read(struct file *f, char
__user
> *buf, mutex_lock(&fifo->read_lock);
> ret = wait_event_interruptible_timeout(fifo->read_queue,
> ioread32(fifo->base_addr + XLLF_RDFO_OFFSET),
> - (read_timeout >= 0) ?
> - msecs_to_jiffies(read_timeout) :
> - MAX_SCHEDULE_TIMEOUT);
> + read_timeout);
>
> if (ret <= 0) {
> if (ret == 0) {
> @@ -528,9 +526,7 @@ static ssize_t axis_fifo_write(struct file *f, const
char
> __user *buf, ret = wait_event_interruptible_timeout(fifo->write_queue,
> ioread32(fifo->base_addr + XLLF_TDFV_OFFSET)
>
> >= words_to_write,
What is this? You haven't yet compiled your patch.
Any further problems with enabling axis-fifo as a module?
Fabio
>
> - (write_timeout >= 0) ?
> - msecs_to_jiffies(write_timeout) :
> - MAX_SCHEDULE_TIMEOUT);
> + write_timeout);
>
> if (ret <= 0) {
> if (ret == 0) {
> @@ -815,6 +811,16 @@ static int axis_fifo_probe(struct platform_device
*pdev)
> char *device_name;
> int rc = 0; /* error return value */
>
> + if (read_timeout >= 0)
> + read_timeout = msecs_to_jiffies(read_timeout);
> + else
> + read_timeout = MAX_SCHEDULE_TIMEOUT;
> +
> + if (write_timeout >= 0)
> + write_timeout = msecs_to_jiffies(write_timeout);
> + else
> + write_timeout = MAX_SCHEDULE_TIMEOUT;
> +
> /* ----------------------------
> * init wrapper device
> * ----------------------------
> --
> 2.34.1
Powered by blists - more mailing lists