* [PATCH] proc: use the same treatment to check proc_lseek as ones for proc_read_iter et.al.
@ 2025-06-07 2:13 wangzijie
2025-08-25 12:30 ` Łukasz P. Michalik
0 siblings, 1 reply; 3+ messages in thread
From: wangzijie @ 2025-06-07 2:13 UTC (permalink / raw)
To: viro, akpm, rick.p.edgecombe, ast, adobriyan, kirill.shutemov,
linux-fsdevel
Cc: wangzijie
Check pde->proc_ops->proc_lseek directly may cause UAF in rmmod scenario. It's a gap
in proc_reg_open() after commit 654b33ada4ab("proc: fix UAF in proc_get_inode()").
Followed by AI Viro's suggestion, fix it in same manner.
Signed-off-by: wangzijie <wangzijie1@honor.com>
---
fs/proc/generic.c | 2 ++
fs/proc/inode.c | 2 +-
fs/proc/internal.h | 5 +++++
include/linux/proc_fs.h | 1 +
4 files changed, 9 insertions(+), 1 deletion(-)
diff --git a/fs/proc/generic.c b/fs/proc/generic.c
index a3e22803c..e0e50914a 100644
--- a/fs/proc/generic.c
+++ b/fs/proc/generic.c
@@ -569,6 +569,8 @@ static void pde_set_flags(struct proc_dir_entry *pde)
if (pde->proc_ops->proc_compat_ioctl)
pde->flags |= PROC_ENTRY_proc_compat_ioctl;
#endif
+ if (pde->proc_ops->proc_lseek)
+ pde->flags |= PROC_ENTRY_proc_lseek;
}
struct proc_dir_entry *proc_create_data(const char *name, umode_t mode,
diff --git a/fs/proc/inode.c b/fs/proc/inode.c
index a3eb3b740..73074b9c7 100644
--- a/fs/proc/inode.c
+++ b/fs/proc/inode.c
@@ -473,7 +473,7 @@ static int proc_reg_open(struct inode *inode, struct file *file)
typeof_member(struct proc_ops, proc_open) open;
struct pde_opener *pdeo;
- if (!pde->proc_ops->proc_lseek)
+ if (!pde_has_proc_lseek(pde))
file->f_mode &= ~FMODE_LSEEK;
if (pde_is_permanent(pde)) {
diff --git a/fs/proc/internal.h b/fs/proc/internal.h
index 96122e91c..3d48ffe72 100644
--- a/fs/proc/internal.h
+++ b/fs/proc/internal.h
@@ -99,6 +99,11 @@ static inline bool pde_has_proc_compat_ioctl(const struct proc_dir_entry *pde)
#endif
}
+static inline bool pde_has_proc_lseek(const struct proc_dir_entry *pde)
+{
+ return pde->flags & PROC_ENTRY_proc_lseek;
+}
+
extern struct kmem_cache *proc_dir_entry_cache;
void pde_free(struct proc_dir_entry *pde);
diff --git a/include/linux/proc_fs.h b/include/linux/proc_fs.h
index ea62201c7..703d0c76c 100644
--- a/include/linux/proc_fs.h
+++ b/include/linux/proc_fs.h
@@ -27,6 +27,7 @@ enum {
PROC_ENTRY_proc_read_iter = 1U << 1,
PROC_ENTRY_proc_compat_ioctl = 1U << 2,
+ PROC_ENTRY_proc_lseek = 1U << 3,
};
struct proc_ops {
--
2.25.1
^ permalink raw reply related [flat|nested] 3+ messages in thread
* Re: [PATCH] proc: use the same treatment to check proc_lseek as ones for proc_read_iter et.al.
2025-06-07 2:13 [PATCH] proc: use the same treatment to check proc_lseek as ones for proc_read_iter et.al wangzijie
@ 2025-08-25 12:30 ` Łukasz P. Michalik
2025-08-25 12:47 ` wangzijie
0 siblings, 1 reply; 3+ messages in thread
From: Łukasz P. Michalik @ 2025-08-25 12:30 UTC (permalink / raw)
To: wangzijie1
Cc: adobriyan, akpm, ast, kirill.shutemov, linux-fsdevel,
rick.p.edgecombe, viro
[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain; charset=./reply, Size: 467 bytes --]
Hello!
This change makes /proc/net/dev a non-seekable file. There is
software that depends on it to be seekable (1), surprisingly some
other files in /proc still work fine after this change.
Minimal C testcase:
==
#include <fcntl.h>
#include <stdio.h>
#include <unistd.h>
int main() {
int fd = open("/proc/net/dev", 0);
off_t o = lseek(fd, 0, SEEK_SET);
if (o == -1) {
perror("lseek");
}
}
==
1) https://github.com/scottchiefbaker/dool/issues/111
--
ŁPM
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: [PATCH] proc: use the same treatment to check proc_lseek as ones for proc_read_iter et.al.
2025-08-25 12:30 ` Łukasz P. Michalik
@ 2025-08-25 12:47 ` wangzijie
0 siblings, 0 replies; 3+ messages in thread
From: wangzijie @ 2025-08-25 12:47 UTC (permalink / raw)
To: lpmichalik
Cc: adobriyan, akpm, ast, kirill.shutemov, linux-fsdevel,
rick.p.edgecombe, viro, wangzijie1
> Hello!
>
> This change makes /proc/net/dev a non-seekable file. There is
> software that depends on it to be seekable (1), surprisingly some
> other files in /proc still work fine after this change.
>
> Minimal C testcase:
> ==
> #include <fcntl.h>
> #include <stdio.h>
> #include <unistd.h>
>
> int main() {
> int fd = open("/proc/net/dev", 0);
> off_t o = lseek(fd, 0, SEEK_SET);
> if (o == -1) {
> perror("lseek");
> }
> }
> ==
>
> 1) https://github.com/scottchiefbaker/dool/issues/111
>
> --
> ŁPM
Thanks for your testcase, I have submitted a patch to fix this:
https://lore.kernel.org/all/20250821105806.1453833-1-wangzijie1@honor.com
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2025-08-25 12:47 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2025-06-07 2:13 [PATCH] proc: use the same treatment to check proc_lseek as ones for proc_read_iter et.al wangzijie
2025-08-25 12:30 ` Łukasz P. Michalik
2025-08-25 12:47 ` wangzijie
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).