[source] Revert "mac80211: remove ath10k delayed initialization hack"
LEDE Commits
lede-commits at lists.infradead.org
Fri Nov 4 04:59:52 PDT 2016
nbd pushed a commit to source.git, branch master:
https://git.lede-project.org/6aa07b82027fa071ebcecf5fe03ecaeca7ec39b2
commit 6aa07b82027fa071ebcecf5fe03ecaeca7ec39b2
Author: Felix Fietkau <nbd at nbd.name>
AuthorDate: Fri Nov 4 12:58:14 2016 +0100
Revert "mac80211: remove ath10k delayed initialization hack"
This reverts commit efd9dec3195c603289ebb552f8650da9dd869828.
ath10k can take a long time to probe, long enough for netifd to fail to
initialize already configured wireless devices
Signed-off-by: Felix Fietkau <nbd at nbd.name>
---
.../921-ath10k_init_devices_synchronously.patch | 33 ++++++++++++++++++++++
1 file changed, 33 insertions(+)
diff --git a/package/kernel/mac80211/patches/921-ath10k_init_devices_synchronously.patch b/package/kernel/mac80211/patches/921-ath10k_init_devices_synchronously.patch
new file mode 100644
index 0000000..e417eb8
--- /dev/null
+++ b/package/kernel/mac80211/patches/921-ath10k_init_devices_synchronously.patch
@@ -0,0 +1,33 @@
+From: Sven Eckelmann <sven at open-mesh.com>
+Date: Tue, 18 Nov 2014 12:29:28 +0100
+Subject: [PATCH] ath10k: Don't initialize devices asynchronously
+
+OpenWrt requires all PHYs to be initialized to create the configuration files
+during bootup. ath10k violates this because it delays the creation of the PHY
+to a not well defined point in the future.
+
+Forcing the work to be done immediately works around this problem but may also
+delay the boot when firmware images cannot be found.
+
+Signed-off-by: Sven Eckelmann <sven at open-mesh.com>
+---
+
+--- a/drivers/net/wireless/ath/ath10k/core.c
++++ b/drivers/net/wireless/ath/ath10k/core.c
+@@ -2220,6 +2220,16 @@ int ath10k_core_register(struct ath10k *
+ ar->chip_id = chip_id;
+ queue_work(ar->workqueue, &ar->register_work);
+
++ /* OpenWrt requires all PHYs to be initialized to create the
++ * configuration files during bootup. ath10k violates this
++ * because it delays the creation of the PHY to a not well defined
++ * point in the future.
++ *
++ * Forcing the work to be done immediately works around this problem
++ * but may also delay the boot when firmware images cannot be found.
++ */
++ flush_workqueue(ar->workqueue);
++
+ return 0;
+ }
+ EXPORT_SYMBOL(ath10k_core_register);
More information about the lede-commits
mailing list