lkml.org 
[lkml]   [2017]   [Jun]   [22]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
Patch in this message
/
In order to make request_*muxed_region() behave more like
mutex_lock(), a possible failure case needs to be eliminated.
When drivers do not properly share the same I/O region, e.g.
one is using request_region() and the other is using
request_muxed_region(), the kernel didn't warn the user about it.
This change modifies IORESOURCE_MUXED behaviour so it always
goes to sleep waiting for the resuorce to be freed and the
inconsistent resource flag usage is logged with KERN_ERR.

v2: Fixed checkpatch.pl warnings and extended the comment
about request_declared_muxed_region.

Signed-off-by: Zoltán Böszörményi <zboszor@pr.hu>
---
kernel/resource.c | 6 +++++-
1 file changed, 5 insertions(+), 1 deletion(-)

diff --git a/kernel/resource.c b/kernel/resource.c
index 2be7029..5df2731 100644
--- a/kernel/resource.c
+++ b/kernel/resource.c
@@ -1125,6 +1125,7 @@ resource_size_t resource_alignment(struct resource *res)
*
* request_declared_muxed_region creates a new shared busy region
* described in an existing resource descriptor.
+ * It only returns if it succeeded.
*
* release_region releases a matching busy region.
* The region is only freed if it was allocated.
@@ -1191,7 +1192,10 @@ struct resource *__request_declared_region(struct resource *parent,
continue;
}
}
- if (conflict->flags & flags & IORESOURCE_MUXED) {
+ if (flags & IORESOURCE_MUXED) {
+ if (!(conflict->flags & IORESOURCE_MUXED))
+ pr_err("Resource conflict between muxed \"%s\" and non-muxed \"%s\" I/O regions!\n",
+ res->name, conflict->name);
add_wait_queue(&muxed_resource_wait, &wait);
write_unlock(&resource_lock);
set_current_state(TASK_UNINTERRUPTIBLE);
--
2.9.4
\
 
 \ /
  Last update: 2017-06-22 20:03    [W:0.351 / U:0.120 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site