On Sun, Jan 23, 2022 at 10:08:37AM +0000, Hao Lee wrote:
> It's not a standard approach that use __get_free_page() to alloc path
> buffer directly. We'd better use kmalloc and PATH_MAX.
> --- a/fs/proc/base.c
> +++ b/fs/proc/base.c
> @@ -1764,25 +1764,26 @@ static const char *proc_pid_get_link(struct dentry *dentry,
>
> static int do_proc_readlink(struct path *path, char __user *buffer, int buflen)
> {
> - char *tmp = (char *)__get_free_page(GFP_KERNEL);
> + char *buf = NULL;
I'd rather not rename anything but keep it minimal.
From: Hao Lee <[email protected]>
It's not a standard approach that use __get_free_page() to alloc path
buffer directly. We'd better use kmalloc and PATH_MAX.
PAGE_SIZE is different on different archs. An unlinked file
with very long canonical pathname will readlink differently
because "(deleted)" eats into a buffer. --adobriyan
Signed-off-by: Hao Lee <[email protected]>
Signed-off-by: Alexey Dobriyan <[email protected]>
---
fs/proc/base.c | 8 ++++----
1 file changed, 4 insertions(+), 4 deletions(-)
--- a/fs/proc/base.c
+++ b/fs/proc/base.c
@@ -1764,25 +1764,25 @@ static const char *proc_pid_get_link(struct dentry *dentry,
static int do_proc_readlink(struct path *path, char __user *buffer, int buflen)
{
- char *tmp = (char *)__get_free_page(GFP_KERNEL);
+ char *tmp = (char *)kmalloc(PATH_MAX, GFP_KERNEL);
char *pathname;
int len;
if (!tmp)
return -ENOMEM;
- pathname = d_path(path, tmp, PAGE_SIZE);
+ pathname = d_path(path, tmp, PATH_MAX);
len = PTR_ERR(pathname);
if (IS_ERR(pathname))
goto out;
- len = tmp + PAGE_SIZE - 1 - pathname;
+ len = tmp + PATH_MAX - 1 - pathname;
if (len > buflen)
len = buflen;
if (copy_to_user(buffer, pathname, len))
len = -EFAULT;
out:
- free_page((unsigned long)tmp);
+ kfree(tmp);
return len;
}