mac80211: backport from trunk r48782 + required kernel patches and mt76, mwlwifi
[15.05/openwrt.git] / package / kernel / mac80211 / patches / 355-b43-fix-support-for-14e4-4321-PCI-dev-with-BCM4321-c.patch
diff --git a/package/kernel/mac80211/patches/355-b43-fix-support-for-14e4-4321-PCI-dev-with-BCM4321-c.patch b/package/kernel/mac80211/patches/355-b43-fix-support-for-14e4-4321-PCI-dev-with-BCM4321-c.patch
deleted file mode 100644 (file)
index 4ecef3b..0000000
+++ /dev/null
@@ -1,32 +0,0 @@
-From: =?UTF-8?q?Rafa=C5=82=20Mi=C5=82ecki?= <zajec5@gmail.com>
-Date: Sat, 6 Jun 2015 22:45:59 +0200
-Subject: [PATCH] b43: fix support for 14e4:4321 PCI dev with BCM4321 chipset
-MIME-Version: 1.0
-Content-Type: text/plain; charset=UTF-8
-Content-Transfer-Encoding: 8bit
-
-It seems Broadcom released two devices with conflicting device id. There
-are for sure 14e4:4321 PCI devices with BCM4321 (N-PHY) chipset, they
-can be found in routers, e.g. Netgear WNR834Bv2. However, according to
-Broadcom public sources 0x4321 is also used for 5 GHz BCM4306 (G-PHY).
-It's unsure if they meant PCI device id, or "virtual" id (from SPROM).
-To distinguish these devices lets check PHY type (G vs. N).
-
-Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
-Cc: <stable@vger.kernel.org> # 3.16+
-Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
----
-
---- a/drivers/net/wireless/b43/main.c
-+++ b/drivers/net/wireless/b43/main.c
-@@ -5365,6 +5365,10 @@ static void b43_supported_bands(struct b
-               *have_5ghz_phy = true;
-               return;
-       case 0x4321: /* BCM4306 */
-+              /* There are 14e4:4321 PCI devs with 2.4 GHz BCM4321 (N-PHY) */
-+              if (dev->phy.type != B43_PHYTYPE_G)
-+                      break;
-+              /* fall through */
-       case 0x4313: /* BCM4311 */
-       case 0x431a: /* BCM4318 */
-       case 0x432a: /* BCM4321 */