Add EXPORT_SYMBOL() to mark freeze_timeout_msecs symbol as visible.

freeze_timeout_msecs defines the timeout for try_to_freeze_tasks. If
tasks cannot be frozen within this timeout, try_to_freeze_tasks fails,
and causes suspend or hibernation to fail.

For example, a device driver may need to perform certain operations
during the freezing phase. In some cases, these operations can take
longer than freeze_timeout_msecs to complete. To allow such drivers to
limit their work appropriately and transition to the frozen state before
the timeout expires, export freeze_timeout_msecs so that it can be
referenced by other kernel components.

Signed-off-by: Dongbum Kim <dongbum.kim@lge.com>
---
 kernel/power/process.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/kernel/power/process.c b/kernel/power/process.c
index dc0dfc349f22..5ddfd645b522 100644
--- a/kernel/power/process.c
+++ b/kernel/power/process.c
@@ -25,6 +25,8 @@
  */
 unsigned int __read_mostly freeze_timeout_msecs = 20 * MSEC_PER_SEC;
 
+EXPORT_SYMBOL(freeze_timeout_msecs);
+
 static int try_to_freeze_tasks(bool user_only)
 {
 	const char *what = user_only ? "user space processes" :
-- 
2.43.0

On Thu, Aug 20, 2026 at 12:44:45PM +0900, Dongbum Kim wrote:
> Add EXPORT_SYMBOL() to mark freeze_timeout_msecs symbol as visible.
> 
> freeze_timeout_msecs defines the timeout for try_to_freeze_tasks. If
> tasks cannot be frozen within this timeout, try_to_freeze_tasks fails,
> and causes suspend or hibernation to fail.
> 
> For example, a device driver may need to perform certain operations

I think you really should know that:

 a) exports without users are a complete no-go, and sending them is
    considered a major offence
 b) even with users, exporting data (variables) is generally a bad idea
 c) non-GPL exports for new functionality are almost always a bad idea

if not please take this up with your corporate and technical leadership
to ensure the Linux training standards at LG are up to date.

On Mon, Aug 24 2026, Christoph Hellwig wrote:
> ...

Hello,

Thanks for the feedback.
I agree that exporting freeze_timeout_msecs variable is not right way.

The real issue is that the driver thread cannot be frozen until timeout
expired for try_to_freeze_tasks. I think freeze_timeout_msecs value is
necessary for all module, including non-GPL modules, to prevent to fail
try_to_freeze_tasks.

I will reconsider the approach.

Thanks,
Dongbum.