[PATCH] memory: Make memory_region_readd_subregion() properly handle mapped aliases

[PATCH] memory: Make memory_region_readd_subregion() properly handle mapped aliases

[David Hildenbrand]

memory_region_readd_subregion() wants to readd a region by first
removing it and then readding it. For readding, it doesn't use one of
the memory_region_add_*() variants, which is why fail to re-increment the
mr->mapped_via_alias counters, resulting in the
assert(alias->mapped_via_alias >= 0) in memory_region_del_subregion()
triggering the next time we call memory_region_readd_subregion().

Fix it by using memory_region_add_subregion_common() for readding the
region.

Reported-by: Niek Linnenbank <nieklinnenbank@gmail.com>
Fixes: 5ead62185d23 ("memory: Make memory_region_is_mapped() succeed when mapped via an alias")
Tested-by: Niek Linnenbank <nieklinnenbank@gmail.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Peter Xu <peterx@redhat.com>
Cc: "Philippe Mathieu-Daudé" <f4bug@amsat.org>
Signed-off-by: David Hildenbrand <david@redhat.com>
---
 softmmu/memory.c | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

diff --git a/softmmu/memory.c b/softmmu/memory.c
index 678dc62f06..76bb1677fe 100644
--- a/softmmu/memory.c
+++ b/softmmu/memory.c
@@ -2626,8 +2626,7 @@ static void memory_region_readd_subregion(MemoryRegion *mr)
         memory_region_transaction_begin();
         memory_region_ref(mr);
         memory_region_del_subregion(container, mr);
-        mr->container = container;
-        memory_region_update_container_subregions(mr);
+        memory_region_add_subregion_common(container, mr->addr, mr);
         memory_region_unref(mr);
         memory_region_transaction_commit();
     }
-- 
2.34.1

Re: [PATCH] memory: Make memory_region_readd_subregion() properly handle mapped aliases

[Paolo Bonzini]

On 2/1/22 11:09, David Hildenbrand wrote:
> memory_region_readd_subregion() wants to readd a region by first
> removing it and then readding it. For readding, it doesn't use one of
> the memory_region_add_*() variants, which is why fail to re-increment the
> mr->mapped_via_alias counters, resulting in the
> assert(alias->mapped_via_alias >= 0) in memory_region_del_subregion()
> triggering the next time we call memory_region_readd_subregion().
> 
> Fix it by using memory_region_add_subregion_common() for readding the
> region.
> 
> Reported-by: Niek Linnenbank <nieklinnenbank@gmail.com>
> Fixes: 5ead62185d23 ("memory: Make memory_region_is_mapped() succeed when mapped via an alias")
> Tested-by: Niek Linnenbank <nieklinnenbank@gmail.com>
> Cc: Paolo Bonzini <pbonzini@redhat.com>
> Cc: Peter Xu <peterx@redhat.com>
> Cc: "Philippe Mathieu-Daudé" <f4bug@amsat.org>
> Signed-off-by: David Hildenbrand <david@redhat.com>
> ---
>   softmmu/memory.c | 3 +--
>   1 file changed, 1 insertion(+), 2 deletions(-)
> 
> diff --git a/softmmu/memory.c b/softmmu/memory.c
> index 678dc62f06..76bb1677fe 100644
> --- a/softmmu/memory.c
> +++ b/softmmu/memory.c
> @@ -2626,8 +2626,7 @@ static void memory_region_readd_subregion(MemoryRegion *mr)
>           memory_region_transaction_begin();
>           memory_region_ref(mr);
>           memory_region_del_subregion(container, mr);
> -        mr->container = container;
> -        memory_region_update_container_subregions(mr);
> +        memory_region_add_subregion_common(container, mr->addr, mr);
>           memory_region_unref(mr);
>           memory_region_transaction_commit();
>       }

Queued, thanks.

Paolo

Re: [PATCH] memory: Make memory_region_readd_subregion() properly handle mapped aliases

[Niek Linnenbank]

[-- Attachment #1: Type: text/plain, Size: 2039 bytes --]

Hi Paolo,

Thanks for queing this patch. I did not yet see it appear in master. Do you
know when we can expect to see it?

Regards,
Niek

On Tue, Feb 1, 2022 at 11:51 AM Paolo Bonzini <pbonzini@redhat.com> wrote:

> On 2/1/22 11:09, David Hildenbrand wrote:
> > memory_region_readd_subregion() wants to readd a region by first
> > removing it and then readding it. For readding, it doesn't use one of
> > the memory_region_add_*() variants, which is why fail to re-increment the
> > mr->mapped_via_alias counters, resulting in the
> > assert(alias->mapped_via_alias >= 0) in memory_region_del_subregion()
> > triggering the next time we call memory_region_readd_subregion().
> >
> > Fix it by using memory_region_add_subregion_common() for readding the
> > region.
> >
> > Reported-by: Niek Linnenbank <nieklinnenbank@gmail.com>
> > Fixes: 5ead62185d23 ("memory: Make memory_region_is_mapped() succeed
> when mapped via an alias")
> > Tested-by: Niek Linnenbank <nieklinnenbank@gmail.com>
> > Cc: Paolo Bonzini <pbonzini@redhat.com>
> > Cc: Peter Xu <peterx@redhat.com>
> > Cc: "Philippe Mathieu-Daudé" <f4bug@amsat.org>
> > Signed-off-by: David Hildenbrand <david@redhat.com>
> > ---
> >   softmmu/memory.c | 3 +--
> >   1 file changed, 1 insertion(+), 2 deletions(-)
> >
> > diff --git a/softmmu/memory.c b/softmmu/memory.c
> > index 678dc62f06..76bb1677fe 100644
> > --- a/softmmu/memory.c
> > +++ b/softmmu/memory.c
> > @@ -2626,8 +2626,7 @@ static void
> memory_region_readd_subregion(MemoryRegion *mr)
> >           memory_region_transaction_begin();
> >           memory_region_ref(mr);
> >           memory_region_del_subregion(container, mr);
> > -        mr->container = container;
> > -        memory_region_update_container_subregions(mr);
> > +        memory_region_add_subregion_common(container, mr->addr, mr);
> >           memory_region_unref(mr);
> >           memory_region_transaction_commit();
> >       }
>
> Queued, thanks.
>
> Paolo
>


-- 
Niek Linnenbank

[-- Attachment #2: Type: text/html, Size: 3131 bytes --]

Re: [PATCH] memory: Make memory_region_readd_subregion() properly handle mapped aliases

[Niek Linnenbank]

[-- Attachment #1: Type: text/plain, Size: 2803 bytes --]

Hi Paolo, Peter,

With the 7.0 release approaching, just a friendly reminder that the patch
in this thread is not yet in master
(2058fdbe81e2985c226a026851dd26b146d3395c).
It's currently preventing the orangepi-pc board to boot SD card images from
U-Boot, terminating with an assert:
   https://www.mail-archive.com/qemu-devel@nongnu.org/msg865382.html

If it helps, I did a full build with David's patch and re-ran all the
boot_linux_console.py tests and all passed OK.

Would it be possible to get this in before the upcoming release?

Thanks in advance,
Niek

On Mon, Feb 28, 2022 at 9:03 PM Niek Linnenbank <nieklinnenbank@gmail.com>
wrote:

> Hi Paolo,
>
> Thanks for queing this patch. I did not yet see it appear in master. Do
> you know when we can expect to see it?
>
> Regards,
> Niek
>
> On Tue, Feb 1, 2022 at 11:51 AM Paolo Bonzini <pbonzini@redhat.com> wrote:
>
>> On 2/1/22 11:09, David Hildenbrand wrote:
>> > memory_region_readd_subregion() wants to readd a region by first
>> > removing it and then readding it. For readding, it doesn't use one of
>> > the memory_region_add_*() variants, which is why fail to re-increment
>> the
>> > mr->mapped_via_alias counters, resulting in the
>> > assert(alias->mapped_via_alias >= 0) in memory_region_del_subregion()
>> > triggering the next time we call memory_region_readd_subregion().
>> >
>> > Fix it by using memory_region_add_subregion_common() for readding the
>> > region.
>> >
>> > Reported-by: Niek Linnenbank <nieklinnenbank@gmail.com>
>> > Fixes: 5ead62185d23 ("memory: Make memory_region_is_mapped() succeed
>> when mapped via an alias")
>> > Tested-by: Niek Linnenbank <nieklinnenbank@gmail.com>
>> > Cc: Paolo Bonzini <pbonzini@redhat.com>
>> > Cc: Peter Xu <peterx@redhat.com>
>> > Cc: "Philippe Mathieu-Daudé" <f4bug@amsat.org>
>> > Signed-off-by: David Hildenbrand <david@redhat.com>
>> > ---
>> >   softmmu/memory.c | 3 +--
>> >   1 file changed, 1 insertion(+), 2 deletions(-)
>> >
>> > diff --git a/softmmu/memory.c b/softmmu/memory.c
>> > index 678dc62f06..76bb1677fe 100644
>> > --- a/softmmu/memory.c
>> > +++ b/softmmu/memory.c
>> > @@ -2626,8 +2626,7 @@ static void
>> memory_region_readd_subregion(MemoryRegion *mr)
>> >           memory_region_transaction_begin();
>> >           memory_region_ref(mr);
>> >           memory_region_del_subregion(container, mr);
>> > -        mr->container = container;
>> > -        memory_region_update_container_subregions(mr);
>> > +        memory_region_add_subregion_common(container, mr->addr, mr);
>> >           memory_region_unref(mr);
>> >           memory_region_transaction_commit();
>> >       }
>>
>> Queued, thanks.
>>
>> Paolo
>>
>
>
> --
> Niek Linnenbank
>
>

-- 
Niek Linnenbank

[-- Attachment #2: Type: text/html, Size: 4403 bytes --]

Re: [PATCH] memory: Make memory_region_readd_subregion() properly handle mapped aliases

[Philippe Mathieu-Daudé via]

On 2/1/22 11:09, David Hildenbrand wrote:
> memory_region_readd_subregion() wants to readd a region by first
> removing it and then readding it. For readding, it doesn't use one of
> the memory_region_add_*() variants, which is why fail to re-increment the
> mr->mapped_via_alias counters, resulting in the
> assert(alias->mapped_via_alias >= 0) in memory_region_del_subregion()
> triggering the next time we call memory_region_readd_subregion().
> 
> Fix it by using memory_region_add_subregion_common() for readding the
> region.
> 
> Reported-by: Niek Linnenbank <nieklinnenbank@gmail.com>
> Fixes: 5ead62185d23 ("memory: Make memory_region_is_mapped() succeed when mapped via an alias")
> Tested-by: Niek Linnenbank <nieklinnenbank@gmail.com>
> Cc: Paolo Bonzini <pbonzini@redhat.com>
> Cc: Peter Xu <peterx@redhat.com>
> Cc: "Philippe Mathieu-Daudé" <f4bug@amsat.org>

Tested-by: Niek Linnenbank <nieklinnenbank@gmail.com>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>

> Signed-off-by: David Hildenbrand <david@redhat.com>
> ---
>  softmmu/memory.c | 3 +--
>  1 file changed, 1 insertion(+), 2 deletions(-)
> 
> diff --git a/softmmu/memory.c b/softmmu/memory.c
> index 678dc62f06..76bb1677fe 100644
> --- a/softmmu/memory.c
> +++ b/softmmu/memory.c
> @@ -2626,8 +2626,7 @@ static void memory_region_readd_subregion(MemoryRegion *mr)
>          memory_region_transaction_begin();
>          memory_region_ref(mr);
>          memory_region_del_subregion(container, mr);
> -        mr->container = container;
> -        memory_region_update_container_subregions(mr);
> +        memory_region_add_subregion_common(container, mr->addr, mr);
>          memory_region_unref(mr);
>          memory_region_transaction_commit();
>      }

Re: [PATCH] memory: Make memory_region_readd_subregion() properly handle mapped aliases

[Philippe Mathieu-Daudé]

On 1/2/22 11:09, David Hildenbrand wrote:
> memory_region_readd_subregion() wants to readd a region by first
> removing it and then readding it. For readding, it doesn't use one of
> the memory_region_add_*() variants, which is why fail to re-increment the
> mr->mapped_via_alias counters, resulting in the
> assert(alias->mapped_via_alias >= 0) in memory_region_del_subregion()
> triggering the next time we call memory_region_readd_subregion().
> 
> Fix it by using memory_region_add_subregion_common() for readding the
> region.
> 
> Reported-by: Niek Linnenbank <nieklinnenbank@gmail.com>
> Fixes: 5ead62185d23 ("memory: Make memory_region_is_mapped() succeed when mapped via an alias")
> Tested-by: Niek Linnenbank <nieklinnenbank@gmail.com>
> Cc: Paolo Bonzini <pbonzini@redhat.com>
> Cc: Peter Xu <peterx@redhat.com>
> Cc: "Philippe Mathieu-Daudé" <f4bug@amsat.org>
> Signed-off-by: David Hildenbrand <david@redhat.com>
> ---
>   softmmu/memory.c | 3 +--
>   1 file changed, 1 insertion(+), 2 deletions(-)

Thanks, queued to mips-fixes.