diff mbox

[Xen-devel,for,4.5] xen/arm: p2m: Correctly initialize cur_offset

Message ID 1412088716-32557-1-git-send-email-julien.grall@linaro.org
State Superseded, archived
Headers show

Commit Message

Julien Grall Sept. 30, 2014, 2:51 p.m. UTC
{~0,} only initializes the first cell of the array to ~0. The other cells
are initialized to 0.

Change the initialization to a loop and, at the same time, do the same
for the mappings.

This is fixing boot after 82985d7 "xen: arm: handle variable p2m levels
in apply_p2m_changes" on platform where the root-level doesn't have
concatenate table (such as the Foundation Model).

---
    This is a bug fix for Xen 4.5. This make Xen booting again on the
    Foundation Model and any platform that doesn't have concatenate
    table for root-level.
---
 xen/arch/arm/p2m.c |   10 ++++++++--
 1 file changed, 8 insertions(+), 2 deletions(-)

Comments

Ian Campbell Oct. 1, 2014, 10:33 a.m. UTC | #1
On Tue, 2014-09-30 at 15:51 +0100, Julien Grall wrote:
> {~0,} only initializes the first cell of the array to ~0. The other cells
> are initialized to 0.
> 
> Change the initialization to a loop and, at the same time, do the same
> for the mappings.

I'm not sure a loop is the best option here vs either { ~0, ~0, ~0, ~0 }
or a memset, but OK.

I don't think mappings is wrong as it is though, any reason to change?

> This is fixing boot after 82985d7 "xen: arm: handle variable p2m levels
> in apply_p2m_changes" on platform where the root-level doesn't have
> concatenate table (such as the Foundation Model).

You forgot your S-o-b.
Julien Grall Oct. 1, 2014, 3 p.m. UTC | #2
On 10/01/2014 11:33 AM, Ian Campbell wrote:
> On Tue, 2014-09-30 at 15:51 +0100, Julien Grall wrote:
>> {~0,} only initializes the first cell of the array to ~0. The other cells
>> are initialized to 0.
>>
>> Change the initialization to a loop and, at the same time, do the same
>> for the mappings.
> 
> I'm not sure a loop is the best option here vs either { ~0, ~0, ~0, ~0 }
> or a memset, but OK.

Thinking a bit more, we effectively need to initialize only the first 2
cells. It's because the P2M can start either at level 0 or level 1.

Anyway I will use { ~0, ~0, ~0, ~0 }.

> I don't think mappings is wrong as it is though, any reason to change?

{ NULL, } initialize the first cell to NULL and the other to 0.

As NULL is equal to (void *)0 it's fine, but I would prefer if we can
get a rid of this construction.

I would prefer if we do { NULL, NULL, NULL, NULL }.

>> This is fixing boot after 82985d7 "xen: arm: handle variable p2m levels
>> in apply_p2m_changes" on platform where the root-level doesn't have
>> concatenate table (such as the Foundation Model).
> 
> You forgot your S-o-b.

Oh right. I will resend a patch
Ian Campbell Oct. 1, 2014, 3:03 p.m. UTC | #3
On Wed, 2014-10-01 at 16:00 +0100, Julien Grall wrote:
> On 10/01/2014 11:33 AM, Ian Campbell wrote:
> > On Tue, 2014-09-30 at 15:51 +0100, Julien Grall wrote:
> >> {~0,} only initializes the first cell of the array to ~0. The other cells
> >> are initialized to 0.
> >>
> >> Change the initialization to a loop and, at the same time, do the same
> >> for the mappings.
> > 
> > I'm not sure a loop is the best option here vs either { ~0, ~0, ~0, ~0 }
> > or a memset, but OK.
> 
> Thinking a bit more, we effectively need to initialize only the first 2
> cells. It's because the P2M can start either at level 0 or level 1.
> 
> Anyway I will use { ~0, ~0, ~0, ~0 }.
> 
> > I don't think mappings is wrong as it is though, any reason to change?
> 
> { NULL, } initialize the first cell to NULL and the other to 0.
> 
> As NULL is equal to (void *)0 it's fine, but I would prefer if we can
> get a rid of this construction.
> 
> I would prefer if we do { NULL, NULL, NULL, NULL }.

OK.

> 
> >> This is fixing boot after 82985d7 "xen: arm: handle variable p2m levels
> >> in apply_p2m_changes" on platform where the root-level doesn't have
> >> concatenate table (such as the Foundation Model).
> > 
> > You forgot your S-o-b.
> 
> Oh right. I will resend a patch
> 
> 
>
diff mbox

Patch

diff --git a/xen/arch/arm/p2m.c b/xen/arch/arm/p2m.c
index 70929fc..3fa7e63 100644
--- a/xen/arch/arm/p2m.c
+++ b/xen/arch/arm/p2m.c
@@ -716,15 +716,21 @@  static int apply_p2m_changes(struct domain *d,
 {
     int rc, ret;
     struct p2m_domain *p2m = &d->arch.p2m;
-    lpae_t *mappings[4] = { NULL, };
+    lpae_t *mappings[4];
     paddr_t addr, orig_maddr = maddr;
     unsigned int level = 0;
     unsigned int cur_root_table = ~0;
-    unsigned int cur_offset[4] = { ~0, };
+    unsigned int cur_offset[4];
     unsigned int count = 0;
     bool_t flush = false;
     bool_t flush_pt;
 
+    for (level = 0; level < 4; level++)
+    {
+        mappings[level] = NULL;
+        cur_offset[level] = ~0;
+    }
+
     /* Some IOMMU don't support coherent PT walk. When the p2m is
      * shared with the CPU, Xen has to make sure that the PT changes have
      * reached the memory