x86/mm: Clean up p2m_finish_type_change return value
PATCH v2 1/1 x86/mm: Clean up p2m_finish_type_change return value
Changed the return value of 1 to 0 so now p2m_finish_type_change returns 0 for success or <0 for error.
The “root” caller of p2m_finish_type_change() is XEN_DMOP_map_mem_type_to_ioreq_server and this does nothing useful with positive values.
---
Changes since V1:
- Update commit message
- Move rc = 0 before out tag.
--- xen/arch/x86/mm/p2m.c | 11 +++-------- 1 file changed, 3 insertions(+), 8 deletions(-)
xen/arch/x86/mm/p2m.c +3 -8
@@ -1188,22 +1188,17 @@ int p2m_finish_type_change(struct domain *d,if ( d->arch.altp2m_eptp[i] !=mfn_x (INVALID_MFN ) ) {struct p2m_domain *altp2m = d->arch.altp2m_p2m[i]; -int rc1;p2m_lock (altp2m); - rc1 =finish_type_change (altp2m, first_gfn, max_nr); + rc =finish_type_change (altp2m, first_gfn, max_nr);p2m_unlock (altp2m); -if ( rc1 <0 ) - { - rc = rc1; +if ( rc <0 ) goto out; - } - - rc |= rc1; } }#endif + rc =0 ; out:p2m_unlock (hostp2m);
Signature · 6 lines
2.17.1 _______________________________________________ Xen-devel mailing list [email protected] https://lists.xenproject.org/mailman/listinfo/xen-devel
Quoted 6 lines
Quoted 1 line
Quoted 1 line · On 17.01.19 at 10:06, <[email protected]> wrote:
Changed the return value of 1 to 0 so now
p2m_finish_type_changereturns 0 for success or <0 for error.
The “root” caller ofp2m_finish_type_change()isXEN_DMOP_map_mem_type_to_ioreq_serverand this does nothing useful with positive values.
By "root" I supposed you mean "only current"? I have to admit that I'm only half way happy with this as the underlying reason. I would expect for there to have been a reason for the return value being this way, and that's what would need to be reasoned against. Furthermore the question then is whether down the call tree the same adjustment shouldn't be made. After all (and possibly just as a first step) why would finish_type_change() return a positive value if its only caller doesn't care?
But anyway, I'm not the maintainer here, so what I say may well be completely ignored.
Jan
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Ping. Is this ok with you, George?
Regards,
Alex
Quoted 38 lines · Alexandru Stefan ISAILA
Changed the return value of 1 to 0 so now
p2m_finish_type_changereturns 0 for success or <0 for error.
The “root” caller ofp2m_finish_type_change()isXEN_DMOP_map_mem_type_to_ioreq_serverand this does nothing useful with positive values.Suggested-by George Dunlap <[email protected]>Signed-off-by Alexandru Isaila <[email protected]>---
Changes since V1:
- Update commit message
- Move rc = 0 before out tag.--- xen/arch/x86/mm/p2m.c | 11 +++-------- 1 file changed, 3 insertions(+), 8 deletions(-)xen/arch/x86/mm/p2m.c +2
-5@@ -1188,22 +1188,17 @@ int p2m_finish_type_change(struct domain *d,if ( d->arch.altp2m_eptp[i] !=mfn_x (INVALID_MFN ) ) {struct p2m_domain *altp2m = d->arch.altp2m_p2m[i]; -int rc1;p2m_lock (altp2m); - rc1 =finish_type_change (altp2m, first_gfn, max_nr); + rc =finish_type_change (altp2m, first_gfn, max_nr);p2m_unlock (altp2m); -if ( rc1 <0 ) - { - rc = rc1; +if ( rc <0 ) goto out;
xen/arch/x86/mm/p2m.c int p2m_finish_type_change(struct domain *d, · line 1204
Quoted 10 lines
xen/arch/x86/mm/p2m.c +1
-3- } - - rc |= rc1; } }#endif + rc =0 ; out:p2m_unlock (hostp2m);
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Quoted 1 line · Alexandru Stefan ISAILA
Ping. Is this ok with you, George?
Sorry -- somehow I thought I'd responded to this, but apparently not.
Quoted 6 lines · Alexandru Stefan ISAILA
Quoted 5 lines
Changed the return value of 1 to 0 so now
p2m_finish_type_changereturns 0 for success or <0 for error.
The “root” caller ofp2m_finish_type_change()isXEN_DMOP_map_mem_type_to_ioreq_serverand this does nothing useful with positive values.
I realize you're just copying what I said in the email where I suggested this, but as Jan pointed out in v1, commit message needs to be more thorough than an off-hand comment during review. I also agree with him that it makes more sense to 'sanitize' ept's recalc_entry() return value in finish_type_change(), rather than at p2m_finish_type_change().
I think it's probably good to get feedback from Paul, who wrote this interface (IIRC) and works with one of its main users (QEMU).
My normal 'template' is "Situation / Problem / Solution": What's the current situation, why is that a problem, and how does this patch fix it.
With that in mind (and also bringing Paul up to speed), I think I'd say something like this:
---
In the case of any errors, finish_type_change() passes values returned from p2m->recalc() up the stack (with some exceptions in the case where an error is expected); this eventually ends up being returned to the XEN_DOMOP_map_mem_type_to_ioreq_server hypercall.
However, on Intel processors (but not on AMD processor), p2m->recalc() can also return '1' as well as '0'. This case is handled very
inconsistently: finish_type_change() will return the value of the final entry it attempts, discarding results for other entries; p2m_finish_type_change() will attempt to accumulate '1's, so that it returns '1' if any of the calls to finish_type_change() returns '1'; and dm_op() will again return '1' only if the very last call to p2m_finish_type_change() returns '1'. The result is that the XEN_DMOP_map_mem_type_to_ioreq_server() hypercall will sometimes return 0 and sometimes return 1 on success, in an unpredictable manner.
The hypercall documentation doesn't mention return values; but it's not clear what the caller could do with the information about whether entries had been changed or not. At the moment it's always 0 on AMD boxes, and *usually* 1 on Intel boxes; so nothing can be relying on a '1' return value for correctness (or if it is, it's broken).
Make the return value on success consistently '0' by only returning 0/-ERROR from finish_type_change(). Also remove the accumulation code from p2m_finish_type_change().
---
Thoughts?
-George
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel