From d88f3b8a427acd1aef7ec3fc4c8e47c7a665d7d0 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Rafa=C5=82=20Mi=C5=82ecki?= Date: Thu, 21 Oct 2021 17:26:56 +0200 Subject: bcm47xx: add kernel 5.10 support MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit It's for *development* only as it doesn't work with lzma-loader due to bigger size. Signed-off-by: Rafał Miłecki --- .../bcm47xx/patches-5.10/300-fork_cacheflush.patch | 21 +++++++++++++++++++++ 1 file changed, 21 insertions(+) create mode 100644 target/linux/bcm47xx/patches-5.10/300-fork_cacheflush.patch (limited to 'target/linux/bcm47xx/patches-5.10/300-fork_cacheflush.patch') diff --git a/target/linux/bcm47xx/patches-5.10/300-fork_cacheflush.patch b/target/linux/bcm47xx/patches-5.10/300-fork_cacheflush.patch new file mode 100644 index 0000000000..daa2c1adf0 --- /dev/null +++ b/target/linux/bcm47xx/patches-5.10/300-fork_cacheflush.patch @@ -0,0 +1,21 @@ +From: Wolfram Joost +Subject: [PATCH] fork_cacheflush + +On ASUS WL-500gP there are many unexpected "Segmentation fault"s that +seem to be caused by a kernel. They can be avoided by: +1) Disabling highpage +2) Using flush_cache_mm in flush_cache_dup_mm + +For details see OpenWrt ticket #2035 https://dev.openwrt.org/ticket/2035 +--- +--- a/arch/mips/include/asm/cacheflush.h ++++ b/arch/mips/include/asm/cacheflush.h +@@ -46,7 +46,7 @@ + extern void (*flush_cache_all)(void); + extern void (*__flush_cache_all)(void); + extern void (*flush_cache_mm)(struct mm_struct *mm); +-#define flush_cache_dup_mm(mm) do { (void) (mm); } while (0) ++#define flush_cache_dup_mm(mm) flush_cache_mm(mm) + extern void (*flush_cache_range)(struct vm_area_struct *vma, + unsigned long start, unsigned long end); + extern void (*flush_cache_page)(struct vm_area_struct *vma, unsigned long page, unsigned long pfn); -- cgit v1.2.3