netdev.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
* [PATCH iproute2] ip xfrm: support 64bit kernel and 32bit userspace
@ 2014-11-28  6:58 roy.qing.li
  2014-11-28  7:08 ` FengYu LeiDian
                   ` (2 more replies)
  0 siblings, 3 replies; 7+ messages in thread
From: roy.qing.li @ 2014-11-28  6:58 UTC (permalink / raw)
  To: netdev

From: Li RongQing <roy.qing.li@gmail.com> 

The size of struct xfrm_userpolicy_info is 168 bytes for 64bit kernel, and
164 bytes for 32bit userspace because of the different alignment. and lead
to "ip xfrm" be unable to work.

add a pad in struct xfrm_userpolicy_info, and enable it by set
KERNEL_64_USERSPACE_32 to y

Signed-off-by: Li RongQing <roy.qing.li@gmail.com>
---
 Makefile             | 4 ++++
 include/linux/xfrm.h | 3 +++
 2 files changed, 7 insertions(+)

diff --git a/Makefile b/Makefile
index 9dbb29f..3cd0d67 100644
--- a/Makefile
+++ b/Makefile
@@ -18,6 +18,10 @@ ifneq ($(SHARED_LIBS),y)
 DEFINES+= -DNO_SHARED_LIBS
 endif
 
+ifeq ($(KERNEL_64_USERSPACE_32),y)
+DEFINES+= -DKNL_64_US_32
+endif
+
 DEFINES+=-DCONFDIR=\"$(CONFDIR)\"
 
 #options for decnet
diff --git a/include/linux/xfrm.h b/include/linux/xfrm.h
index fa2ecb2..009510c 100644
--- a/include/linux/xfrm.h
+++ b/include/linux/xfrm.h
@@ -407,6 +407,9 @@ struct xfrm_userpolicy_info {
 	/* Automatically expand selector to include matching ICMP payloads. */
 #define XFRM_POLICY_ICMP	2
 	__u8				share;
+#ifdef KNL_64_US_32
+	int				pad;
+#endif
 };
 
 struct xfrm_userpolicy_id {
-- 
2.1.0

^ permalink raw reply related	[flat|nested] 7+ messages in thread

* Re: [PATCH iproute2] ip xfrm: support 64bit kernel and 32bit userspace
  2014-11-28  6:58 [PATCH iproute2] ip xfrm: support 64bit kernel and 32bit userspace roy.qing.li
@ 2014-11-28  7:08 ` FengYu LeiDian
  2014-11-28  7:35   ` Li RongQing
  2014-11-28 10:47 ` David Laight
  2014-11-30  1:33 ` Ben Hutchings
  2 siblings, 1 reply; 7+ messages in thread
From: FengYu LeiDian @ 2014-11-28  7:08 UTC (permalink / raw)
  To: roy.qing.li; +Cc: netdev

于 2014年11月28日 14:58, roy.qing.li@gmail.com 写道:
> From: Li RongQing<roy.qing.li@gmail.com>  
> 
> The size of struct xfrm_userpolicy_info is 168 bytes for 64bit kernel, and
> 164 bytes for 32bit userspace because of the different alignment. and lead
> to "ip xfrm" be unable to work.
> 
> add a pad in struct xfrm_userpolicy_info, and enable it by set
> KERNEL_64_USERSPACE_32 to y


It's not easy to fix it like this.

Refer:
http://marc.info/?l=linux-netdev&m=139280143513650&w=2
http://thread.gmane.org/gmane.linux.network/157118/focus=157122

^ permalink raw reply	[flat|nested] 7+ messages in thread

* Re: [PATCH iproute2] ip xfrm: support 64bit kernel and 32bit userspace
  2014-11-28  7:08 ` FengYu LeiDian
@ 2014-11-28  7:35   ` Li RongQing
  2014-11-28  8:07     ` FengYu LeiDian
  0 siblings, 1 reply; 7+ messages in thread
From: Li RongQing @ 2014-11-28  7:35 UTC (permalink / raw)
  To: FengYu LeiDian; +Cc: netdev

On Fri, Nov 28, 2014 at 3:08 PM, FengYu LeiDian
<fengyuleidian0615@gmail.com> wrote:
> 于 2014年11月28日 14:58, roy.qing.li@gmail.com 写道:
>> From: Li RongQing<roy.qing.li@gmail.com>
>>
>> The size of struct xfrm_userpolicy_info is 168 bytes for 64bit kernel, and
>> 164 bytes for 32bit userspace because of the different alignment. and lead
>> to "ip xfrm" be unable to work.
>>
>> add a pad in struct xfrm_userpolicy_info, and enable it by set
>> KERNEL_64_USERSPACE_32 to y
>
>
> It's not easy to fix it like this.
>
> Refer:
> http://marc.info/?l=linux-netdev&m=139280143513650&w=2
> http://thread.gmane.org/gmane.linux.network/157118/focus=157122
>

is it be practical to find all structures which has the issue for
64bit kernel and 32 bit
userspace, and conditionally add a pad for them

-Roy

^ permalink raw reply	[flat|nested] 7+ messages in thread

* Re: [PATCH iproute2] ip xfrm: support 64bit kernel and 32bit userspace
  2014-11-28  7:35   ` Li RongQing
@ 2014-11-28  8:07     ` FengYu LeiDian
  0 siblings, 0 replies; 7+ messages in thread
From: FengYu LeiDian @ 2014-11-28  8:07 UTC (permalink / raw)
  To: Li RongQing; +Cc: netdev

于 2014年11月28日 15:35, Li RongQing 写道:
> On Fri, Nov 28, 2014 at 3:08 PM, FengYu LeiDian
> <fengyuleidian0615@gmail.com>  wrote:
>> >于 2014年11月28日 14:58,roy.qing.li@gmail.com  写道:
>>> >>From: Li RongQing<roy.qing.li@gmail.com>
>>> >>
>>> >>The size of struct xfrm_userpolicy_info is 168 bytes for 64bit kernel, and
>>> >>164 bytes for 32bit userspace because of the different alignment. and lead
>>> >>to "ip xfrm" be unable to work.
>>> >>
>>> >>add a pad in struct xfrm_userpolicy_info, and enable it by set
>>> >>KERNEL_64_USERSPACE_32 to y
>> >
>> >
>> >It's not easy to fix it like this.
>> >
>> >Refer:
>> >http://marc.info/?l=linux-netdev&m=139280143513650&w=2
>> >http://thread.gmane.org/gmane.linux.network/157118/focus=157122
>> >
> is it be practical to find all structures which has the issue for
> 64bit kernel and 32 bit
> userspace, and conditionally add a pad for them

Apparently struct xfrm_userpolicy_info is not the only one,
you could try to set SA as well, it breaks also...
Fix those where it hurts, hehe :) Otherwise, the "compat layer"
fix style is good choice.

^ permalink raw reply	[flat|nested] 7+ messages in thread

* RE: [PATCH iproute2] ip xfrm: support 64bit kernel and 32bit userspace
  2014-11-28  6:58 [PATCH iproute2] ip xfrm: support 64bit kernel and 32bit userspace roy.qing.li
  2014-11-28  7:08 ` FengYu LeiDian
@ 2014-11-28 10:47 ` David Laight
  2014-11-30  1:33 ` Ben Hutchings
  2 siblings, 0 replies; 7+ messages in thread
From: David Laight @ 2014-11-28 10:47 UTC (permalink / raw)
  To: 'roy.qing.li@gmail.com', netdev@vger.kernel.org

From: roy.qing.li@gmail.com
> From: Li RongQing <roy.qing.li@gmail.com>
> 
> The size of struct xfrm_userpolicy_info is 168 bytes for 64bit kernel, and
> 164 bytes for 32bit userspace because of the different alignment. and lead
> to "ip xfrm" be unable to work.
> 
> add a pad in struct xfrm_userpolicy_info, and enable it by set
> KERNEL_64_USERSPACE_32 to y
...
> diff --git a/include/linux/xfrm.h b/include/linux/xfrm.h
> index fa2ecb2..009510c 100644
> --- a/include/linux/xfrm.h
> +++ b/include/linux/xfrm.h
> @@ -407,6 +407,9 @@ struct xfrm_userpolicy_info {
>  	/* Automatically expand selector to include matching ICMP payloads. */
>  #define XFRM_POLICY_ICMP	2
>  	__u8				share;
> +#ifdef KNL_64_US_32
> +	int				pad;
> +#endif
>  };

You could add __attribute__((aligned(8))) to the structure to pad it on
all systems.

Or find the 64bit member and add __attribute__((aligned(4))) to that member
so that the 64bit version doesn't get padded.
That will make the field more expensive to access on systems that can't
do misaligned memory transfers.

	David

^ permalink raw reply	[flat|nested] 7+ messages in thread

* Re: [PATCH iproute2] ip xfrm: support 64bit kernel and 32bit userspace
  2014-11-28  6:58 [PATCH iproute2] ip xfrm: support 64bit kernel and 32bit userspace roy.qing.li
  2014-11-28  7:08 ` FengYu LeiDian
  2014-11-28 10:47 ` David Laight
@ 2014-11-30  1:33 ` Ben Hutchings
  2014-11-30 17:42   ` Stephen Hemminger
  2 siblings, 1 reply; 7+ messages in thread
From: Ben Hutchings @ 2014-11-30  1:33 UTC (permalink / raw)
  To: roy.qing.li; +Cc: netdev

[-- Attachment #1: Type: text/plain, Size: 838 bytes --]

On Fri, 2014-11-28 at 14:58 +0800, roy.qing.li@gmail.com wrote:
> From: Li RongQing <roy.qing.li@gmail.com> 
> 
> The size of struct xfrm_userpolicy_info is 168 bytes for 64bit kernel, and
> 164 bytes for 32bit userspace because of the different alignment.

I think it's specific to x86 as other architectures have the same type
alignments for 32- and 64-bit variants.

> and lead to "ip xfrm" be unable to work.
> 
> add a pad in struct xfrm_userpolicy_info, and enable it by set
> KERNEL_64_USERSPACE_32 to y
[...]

This doesn't make sense.  32-bit compat needs to work like 32-bit native
- no exceptions, no workarounds in userland.  And if you really want to
work around this in userland you'll need a run-time, not compile-time,
check.

Ben.

-- 
Ben Hutchings
The world is coming to an end.	Please log off.

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 811 bytes --]

^ permalink raw reply	[flat|nested] 7+ messages in thread

* Re: [PATCH iproute2] ip xfrm: support 64bit kernel and 32bit userspace
  2014-11-30  1:33 ` Ben Hutchings
@ 2014-11-30 17:42   ` Stephen Hemminger
  0 siblings, 0 replies; 7+ messages in thread
From: Stephen Hemminger @ 2014-11-30 17:42 UTC (permalink / raw)
  To: Ben Hutchings; +Cc: roy.qing.li, netdev

[-- Attachment #1: Type: text/plain, Size: 973 bytes --]

On Sun, 30 Nov 2014 01:33:22 +0000
Ben Hutchings <ben@decadent.org.uk> wrote:

> On Fri, 2014-11-28 at 14:58 +0800, roy.qing.li@gmail.com wrote:
> > From: Li RongQing <roy.qing.li@gmail.com> 
> > 
> > The size of struct xfrm_userpolicy_info is 168 bytes for 64bit kernel, and
> > 164 bytes for 32bit userspace because of the different alignment.
> 
> I think it's specific to x86 as other architectures have the same type
> alignments for 32- and 64-bit variants.
> 
> > and lead to "ip xfrm" be unable to work.
> > 
> > add a pad in struct xfrm_userpolicy_info, and enable it by set
> > KERNEL_64_USERSPACE_32 to y
> [...]
> 
> This doesn't make sense.  32-bit compat needs to work like 32-bit native
> - no exceptions, no workarounds in userland.  And if you really want to
> work around this in userland you'll need a run-time, not compile-time,
> check.

I agree with Ben, this is not a good solution.
It creates a 3rd ABI (32, 64, and 32/64).

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 819 bytes --]

^ permalink raw reply	[flat|nested] 7+ messages in thread

end of thread, other threads:[~2014-11-30 17:42 UTC | newest]

Thread overview: 7+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2014-11-28  6:58 [PATCH iproute2] ip xfrm: support 64bit kernel and 32bit userspace roy.qing.li
2014-11-28  7:08 ` FengYu LeiDian
2014-11-28  7:35   ` Li RongQing
2014-11-28  8:07     ` FengYu LeiDian
2014-11-28 10:47 ` David Laight
2014-11-30  1:33 ` Ben Hutchings
2014-11-30 17:42   ` Stephen Hemminger

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).