Skip to contentSkip to conversation

x86/HVM+shim: fix build when !CONFIG_GRANT_TABLE

PATCH In review Jan Beulich 3 participants 6 messages Dec 1, 2022

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 x86/HVM+shim: fix build when !CONFIG_GRANT_TABLE

The HVM flavor of the hypercall handlers exists only when GRANT_TABLE is enabled, while surrogate shim variants exist only for the purpose of PV. (Also scratch out the Arm variant in that case; what exactly is used in that cell of the new table row doesn't really matter.)

Fixes 8523851dbc49 ("xen/x86: call hypercall handlers via generated macro")
xen/include/hypercall-defs.c +3 -1
@@ -241,8 +241,10 @@ event_channel_op_compat            do
 xen_version                        compat   do       compat   do       do
 console_io                         do       do       do       do       do
 physdev_op_compat                  compat   do       -        -        dep
-#if defined(CONFIG_GRANT_TABLE) || defined(CONFIG_PV_SHIM)
+#if defined(CONFIG_GRANT_TABLE)
 grant_table_op                     compat   do       hvm      hvm      do
+#elif defined(CONFIG_PV_SHIM)
+grant_table_op                     compat   do       -        -        -
 #endif
 vm_assist                          do       do       do       do       do
 update_va_mapping_otherdomain      compat   do       -        -        -
Acked-by on v1 1/1
#1 lore raw
Quoted 8 lines · Jan Beulich

The HVM flavor of the hypercall handlers exists only when GRANT_TABLE is enabled, while surrogate shim variants exist only for the purpose of PV. (Also scratch out the Arm variant in that case; what exactly is used in that cell of the new table row doesn't really matter.)

Fixes 8523851dbc49 ("xen/x86: call hypercall handlers via generated macro")

Thanks for investigating.

Thoughts about inclusion into 4.17?  This is a build time regression vs 4.16.

#2 lore raw

[4.17?] Re: [PATCH] x86/HVM+shim: fix build when !CONFIG_GRANT_TABLE

Quoted 6 lines · Andrew Cooper
Quoted 5 lines · Jan Beulich

The HVM flavor of the hypercall handlers exists only when GRANT_TABLE is enabled, while surrogate shim variants exist only for the purpose of PV. (Also scratch out the Arm variant in that case; what exactly is used in that cell of the new table row doesn't really matter.)

Quoted 7 lines
Quoted 3 lines
Fixes 8523851dbc49 ("xen/x86: call hypercall handlers via generated macro")

Thanks for investigating.

Thanks.

Quoted 2 lines

Thoughts about inclusion into 4.17?  This is a build time regression vs 4.16.

I thought this was odd enough a configuration, but since you ask, let me forward the question to Henry.

Jan

#3 lore raw

RE: [4.17?] Re: [PATCH] x86/HVM+shim: fix build when !CONFIG_GRANT_TABLE

Hi Jan and Andrew,

Quoted 22 lines

-----Original Message-----
From: Jan Beulich <[email protected]>
Subject: [4.17?] Re: [PATCH] x86/HVM+shim: fix build when !CONFIG_GRANT_TABLE

Quoted 2 lines · Andrew Cooper
Quoted 1 line · Jan Beulich

The HVM flavor of the hypercall handlers exists only when GRANT_TABLE

is

Quoted 5 lines
Quoted 5 lines

enabled, while surrogate shim variants exist only for the purpose of PV. (Also scratch out the Arm variant in that case; what exactly is used in that cell of the new table row doesn't really matter.)

Fixes 8523851dbc49 ("xen/x86: call hypercall handlers via generated

macro")

Quoted 6 lines
Quoted 2 lines

Thanks for investigating.

Quoted 7 lines

Thanks.

Quoted 2 lines

Thoughts about inclusion into 4.17?  This is a build time regression vs 4.16.

I thought this was odd enough a configuration, but since you ask, let me forward the question to Henry.

I think a build time regression should be fixed. We cannot assume users will not use this configuration (as Jan pointed out in IRC). So for 4.17:

But if OSSTest is exploded after merging this patch, I would like to request a revert since we don't have too much time left for the due date. Hopefully everyone is fine with that.

Kind regards,
Henry

Quoted 2 lines

Jan

#4 lore raw

Re: [4.17?] Re: [PATCH] x86/HVM+shim: fix build when !CONFIG_GRANT_TABLE

Quoted 9 lines · Henry Wang
Quoted 9 lines
Quoted 3 lines

Thanks for investigating.

Thanks.

Quoted 2 lines

Thoughts about inclusion into 4.17?  This is a build time regression vs 4.16.

I thought this was odd enough a configuration, but since you ask, let me forward the question to Henry.

Quoted 8 lines

I think a build time regression should be fixed. We cannot assume users will not use this configuration (as Jan pointed out in IRC). So for 4.17:

But if OSSTest is exploded after merging this patch, I would like to request a revert since we don't have too much time left for the due date. Hopefully everyone is fine with that.

It is very unlikely that people are going have a configuration like this in production.

But, the 4.17 branch does have Gitlab CI running on it, including randconf tests, which provably do spot the error occasionally.

The (IMO better) justification to take it into 4.17 at this point is to fix a CI failure.

~Andrew

#5 lore raw

RE: [4.17?] Re: [PATCH] x86/HVM+shim: fix build when !CONFIG_GRANT_TABLE

Hi Andrew,

Quoted 24 lines

-----Original Message-----
From: Andrew Cooper <[email protected]>
Subject: Re: [4.17?] Re: [PATCH] x86/HVM+shim: fix build when !CONFIG_GRANT_TABLE

Quoted 9 lines · Henry Wang
Quoted 9 lines
Quoted 3 lines

Thanks for investigating.

Thanks.

Quoted 2 lines

Thoughts about inclusion into 4.17?  This is a build time regression vs 4.16.

I thought this was odd enough a configuration, but since you ask, let me forward the question to Henry.

Quoted 8 lines

I think a build time regression should be fixed. We cannot assume users will not use this configuration (as Jan pointed out in IRC). So for 4.17:

But if OSSTest is exploded after merging this patch, I would like to request a revert since we don't have too much time left for the due date. Hopefully everyone is fine with that.

Quoted 8 lines

It is very unlikely that people are going have a configuration like this in production.

But, the 4.17 branch does have Gitlab CI running on it, including randconf tests, which provably do spot the error occasionally.

The (IMO better) justification to take it into 4.17 at this point is to fix a CI failure.

Good point, my release ack is still valid :)

Kind regards,
Henry

Quoted 2 lines

~Andrew