[lkml]   [2003]   [Aug]   [3]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
Messages in this thread
Patches in this message
SubjectRe: PATCH: 2.4.22-pre7 drivers/i2c/i2c-dev.c user/kernel bug and mem leak

Hi all,

Ten days ago, Robert T. Johnson repported two bugs in 2.4's
drivers/i2c/i2c-dev.c. It also applies to i2c CVS (out of kernel), which
is intended to become 2.4's soon. Being a member of the LM Sensors dev
team, I took a look at the repport. My knowledge is somewhat limited but
I'll do my best to help (unless Greg wants to handle it alone? ;-)).

For the user/kernel bug, I'm not sure I understand how copy_from_user is
supposed to work. If I understand what the proposed patch does, it
simply allocates a second buffer, copy_from_user to that buffer instead
of to the original one, and then copies from that second buffer to the
original one (kernel to kernel). I just don't see how different it is
from what the current code does, as far as user/kernel issues are
concerned. I must be missing something obvious, can someone please bring
me some light?

For the mem leak bug, it's clearly there. I admit the proposed patch
fixes it, but I think there is a better way to fix it. Compare what the
proposed patch does:

--- i2c-dev.c Sun Aug 3 18:24:33 2003
+++ i2c-dev.c.proposed Sun Aug 3 19:13:58 2003
@@ -226,6 +226,7 @@
res = 0;
for( i=0; i<rdwr_arg.nmsgs; i++ )
+ rdwr_pa[i].buf = NULL;
@@ -254,8 +255,9 @@
if (res < 0) {
int j;
- for (j = 0; j < i; ++j)
- kfree(rdwr_pa[j].buf);
+ for (j = 0; j <= i; ++j)
+ if (rdwr_pa[j].buf)
+ kfree(rdwr_pa[j].buf);
return res;
with what I suggest:

--- i2c-dev.c Sun Aug 3 18:24:33 2003
+++ i2c-dev.c.khali Sun Aug 3 19:15:04 2003
@@ -247,8 +247,9 @@
+ kfree(rdwr_pa[i].buf);
res = -EFAULT;
Contrary to the proposed fix, my fix does not slow down the non-faulty
cases. I also believe it will increase the code size by fewer bytes than
the proposed fix (not verified though).

So, what about it?

PS: I really would like to see Frodo Looijaard's address replaced with
the LM Sensors mailing list address <> as
the main I2C contact in MAINTAINERS. Simon Vogl and Frodo Looijaard's
have been doing a really great job, but they do not work actively on I2C
anymore, so they end up forwarding every repport to the list anyway.

Jean Delvare
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to
More majordomo info at
Please read the FAQ at

 \ /
  Last update: 2005-03-22 13:47    [W:0.074 / U:0.028 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site