Hello Arnd,
thanks for picking up this y2038 api issue.
On 09/30/2015 01:26 PM, Arnd Bergmann wrote:
The can subsystem communicates with user space using a bcm_msg_head
header, which contains two timestamps. This is problematic for
multiple reasons:
a) The structure layout is currently incompatible between 64-bit
user space and 32-bit user space, and cannot work in compat
mode (other than x32).
b) The timeval structure layout will change in 32-bit user
space when we fix the y2038 overflow problem by redefining
time_t to 64-bit, making new 32-bit user space incompatible
with the current kernel interface.
Cars last a long time and often use old kernels, so the actual
users of this code are the most likely ones to migrate to y2038
safe user space.
This tries to work around part of the problem by changing the
publicly visible user interface in the header, but not the binary
interface. Fortunately, the values passed around in the structure
are relative times and do not actually suffer from the y2038
overflow, so 32-bit is enough here.
We replace the use of 'struct timeval' with a newly defined
'struct bcm_timeval' that uses the exact same binary layout
as before and that still suffers from problem a) but not problem
b).
The downside of this approach is that any user space program
that currently assigns a timeval structure to these members
rather than writing the tv_sec/tv_usec portions individually
will suffer a compile-time error when built with an updated
kernel header. Fixing this error makes it work fine with old
and new headers though.
I double checked some (more) BCM applications I have access to.
E.g. https://github.com/linux-can/can-tests
When you do a 'git grep ival1' there you get something like
tst-bcm-cycle.c: msg.msg_head.ival1.tv_sec = 1;
tst-bcm-cycle.c: msg.msg_head.ival1.tv_usec = 0;
tst-bcm-cycle.c: msg.msg_head.ival1.tv_sec = 0;
tst-bcm-cycle.c: msg.msg_head.ival1.tv_usec = 0;
tst-bcm-dump.c: msg.msg_head.ival1.tv_sec = timeout / 1000000;
tst-bcm-dump.c: msg.msg_head.ival1.tv_usec = timeout % 1000000;
(..)
So the usual way to assign values to ival1 and ival2 is NOT to assign an
existing struct timeval but to directly assign its tv_[u]sec elements.
I applied your bcm.h changes to my local can-tests tree and it compiles
without any problems - as expected. I don't see any serious drawback with your
idea. I wonder whether developers would ever notice this change ...
We could address problem a) by using '__u32' or 'int' members
rather than 'long', but that would have a more significant
downside in also breaking support for all existing 64-bit user
binaries that might be using this interface, which is likely
not acceptable.
Indeed.
Signed-off-by: Arnd Bergmann <redacted>
Cc: Oliver Hartkopp <redacted>
Thanks for your good suggestion to make the BCM API y2038 proof!
Acked-by: Oliver Hartkopp <redacted>
quoted hunk
Cc: Marc Kleine-Budde <redacted>
Cc: <redacted>
Cc: <redacted>
---
include/uapi/linux/can/bcm.h | 7 ++++++-
net/can/bcm.c | 15 ++++++++++-----
2 files changed, 16 insertions(+), 6 deletions(-)
diff --git a/include/uapi/linux/can/bcm.h b/include/uapi/linux/can/bcm.h
index 89ddb9dc9bdf..7a291dc1ff15 100644
--- a/include/uapi/linux/can/bcm.h
+++ b/include/uapi/linux/can/bcm.h
@@ -47,6 +47,11 @@
#include <linux/types.h>
#include <linux/can.h>
+struct bcm_timeval {
+ long tv_sec;
+ long tv_usec;
+};
+
/**
* struct bcm_msg_head - head of messages to/from the broadcast manager
* @opcode: opcode, see enum below.@@ -62,7 +67,7 @@ struct bcm_msg_head {
__u32 opcode;
__u32 flags;
__u32 count;
- struct timeval ival1, ival2;
+ struct bcm_timeval ival1, ival2;
canid_t can_id;
__u32 nframes;
struct can_frame frames[0];diff --git a/net/can/bcm.c b/net/can/bcm.c
index a1ba6875c2a2..6863310d6973 100644
--- a/net/can/bcm.c
+++ b/net/can/bcm.c
@@ -96,7 +96,7 @@ struct bcm_op {
canid_t can_id;
u32 flags;
41 more quoted lines