PKGBUILDs/alarm/cryptodev-dkms/pageaccess_fix_for_4.6+.patch
2016-08-19 23:52:42 -06:00

52 lines
1.7 KiB
Diff

From f14b4706b0d04988e7e5bc8c4d2aefef9f029d9d Mon Sep 17 00:00:00 2001
From: Michael Weiser <michael.weiser@gmx.de>
Date: Fri, 5 Aug 2016 18:43:55 +0200
Subject: [PATCH] Adjust to recent user page API changes
4.6.0 basically renamed get_user_pages() to get_user_pages_remote() and
introduced a new get_user_pages() that always works on the current
task.[1] Distinguish the two APIs based on kernel version we're
compiling for.
Also, there seems to have been a massive cleansing of
page_cache_release(page) in favour of put_page(page)[2] which was an
alias for put_page(page)[3] since 2.6.0. Before that beginning with
2.4.0 both page_cache_release(page) and put_page(page) have been aliases
for __free_page(page). So using put_page() instead of
page_cache_release(page) will produce identical code for anything after
2.4.0.
[1] https://lkml.org/lkml/2016/2/10/555
[2] https://www.spinics.net/lists/linux-fsdevel/msg95923.html
[3] https://www.spinics.net/lists/linux-fsdevel/msg95922.html
---
zc.c | 9 +++++++--
1 file changed, 7 insertions(+), 2 deletions(-)
diff --git a/zc.c b/zc.c
index 29b0501..a97b49f 100644
--- a/zc.c
+++ b/zc.c
@@ -59,7 +59,12 @@ int __get_userbuf(uint8_t __user *addr, uint32_t len, int write,
}
down_read(&mm->mmap_sem);
- ret = get_user_pages(task, mm,
+#if (LINUX_VERSION_CODE >= KERNEL_VERSION(4, 6, 0))
+ ret = get_user_pages_remote(
+#else
+ ret = get_user_pages(
+#endif
+ task, mm,
(unsigned long)addr, pgcount, write, 0, pg, NULL);
up_read(&mm->mmap_sem);
if (ret != pgcount)
@@ -119,7 +124,7 @@ void release_user_pages(struct csession *ses)
else
ses->readonly_pages--;
- page_cache_release(ses->pages[i]);
+ put_page(ses->pages[i]);
}
ses->used_pages = 0;
}