diff mbox series

physmem: fix qemu_ram_alloc_from_fd size calculation

Message ID 1735850090-329906-1-git-send-email-steven.sistare@oracle.com (mailing list archive)
State New
Headers show
Series physmem: fix qemu_ram_alloc_from_fd size calculation | expand

Commit Message

Steven Sistare Jan. 2, 2025, 8:34 p.m. UTC
qemu_ram_alloc_from_fd allocates space if file_size == 0.  If non-zero,
it uses the existing space and verifies it is large enough, but the
verification was broken when the offset parameter was introduced.  As
a result, a file smaller than offset passes the verification and causes
errors later.  Fix that, and update the error message to include offset.

Peter provides this concise reproducer:

  $ touch ramfile
  $ truncate -s 64M ramfile
  $ ./qemu-system-x86_64 -object memory-backend-file,mem-path=./ramfile,offset=128M,size=128M,id=mem1,prealloc=on
  qemu-system-x86_64: qemu_prealloc_mem: preallocating memory failed: Bad address

With the fix, the error message is:
  qemu-system-x86_64: mem1 backing store size 0x4000000 is too small for 'size' option 0x8000000 plus 'offset' option 0x8000000

Fixes: 4b870dc4d0c0 ("hostmem-file: add offset option")
Signed-off-by: Steve Sistare <steven.sistare@oracle.com>
---
 system/physmem.c | 9 +++++----
 1 file changed, 5 insertions(+), 4 deletions(-)

Comments

Peter Xu Jan. 2, 2025, 9:14 p.m. UTC | #1
On Thu, Jan 02, 2025 at 12:34:50PM -0800, Steve Sistare wrote:
> qemu_ram_alloc_from_fd allocates space if file_size == 0.  If non-zero,
> it uses the existing space and verifies it is large enough, but the
> verification was broken when the offset parameter was introduced.  As
> a result, a file smaller than offset passes the verification and causes
> errors later.  Fix that, and update the error message to include offset.
> 
> Peter provides this concise reproducer:
> 
>   $ touch ramfile
>   $ truncate -s 64M ramfile
>   $ ./qemu-system-x86_64 -object memory-backend-file,mem-path=./ramfile,offset=128M,size=128M,id=mem1,prealloc=on
>   qemu-system-x86_64: qemu_prealloc_mem: preallocating memory failed: Bad address
> 
> With the fix, the error message is:
>   qemu-system-x86_64: mem1 backing store size 0x4000000 is too small for 'size' option 0x8000000 plus 'offset' option 0x8000000
> 

Can have the cc tag here too to be super clear:

Cc: qemu-stable@nongnu.org

> Fixes: 4b870dc4d0c0 ("hostmem-file: add offset option")
> Signed-off-by: Steve Sistare <steven.sistare@oracle.com>

Reviewed-by: Peter Xu <peterx@redhat.com>

Please also feel free to attach this patch as the 1st patch when repost
cpr-transfer, it could make dependency easier no matter how this would
land.

Maybe it'll be still easier to go via a migration pull that collects cpr
transfer as a whole. May depend on whether there'll be a memory API pull
before that..

Thanks,
Steven Sistare Jan. 2, 2025, 9:33 p.m. UTC | #2
On 1/2/2025 4:14 PM, Peter Xu wrote:
> On Thu, Jan 02, 2025 at 12:34:50PM -0800, Steve Sistare wrote:
>> qemu_ram_alloc_from_fd allocates space if file_size == 0.  If non-zero,
>> it uses the existing space and verifies it is large enough, but the
>> verification was broken when the offset parameter was introduced.  As
>> a result, a file smaller than offset passes the verification and causes
>> errors later.  Fix that, and update the error message to include offset.
>>
>> Peter provides this concise reproducer:
>>
>>    $ touch ramfile
>>    $ truncate -s 64M ramfile
>>    $ ./qemu-system-x86_64 -object memory-backend-file,mem-path=./ramfile,offset=128M,size=128M,id=mem1,prealloc=on
>>    qemu-system-x86_64: qemu_prealloc_mem: preallocating memory failed: Bad address
>>
>> With the fix, the error message is:
>>    qemu-system-x86_64: mem1 backing store size 0x4000000 is too small for 'size' option 0x8000000 plus 'offset' option 0x8000000
>>
> 
> Can have the cc tag here too to be super clear:
> 
> Cc: qemu-stable@nongnu.org

Done and resubmitted.

>> Fixes: 4b870dc4d0c0 ("hostmem-file: add offset option")
>> Signed-off-by: Steve Sistare <steven.sistare@oracle.com>
> 
> Reviewed-by: Peter Xu <peterx@redhat.com>
> 
> Please also feel free to attach this patch as the 1st patch when repost
> cpr-transfer, it could make dependency easier no matter how this would
> land.

Will do - steve

> Maybe it'll be still easier to go via a migration pull that collects cpr
> transfer as a whole. May depend on whether there'll be a memory API pull
> before that..
diff mbox series

Patch

diff --git a/system/physmem.c b/system/physmem.c
index c76503a..f01325f 100644
--- a/system/physmem.c
+++ b/system/physmem.c
@@ -1970,10 +1970,11 @@  RAMBlock *qemu_ram_alloc_from_fd(ram_addr_t size, MemoryRegion *mr,
     size = REAL_HOST_PAGE_ALIGN(size);
 
     file_size = get_file_size(fd);
-    if (file_size > offset && file_size < (offset + size)) {
-        error_setg(errp, "backing store size 0x%" PRIx64
-                   " does not match 'size' option 0x" RAM_ADDR_FMT,
-                   file_size, size);
+    if (file_size && file_size < offset + size) {
+        error_setg(errp, "%s backing store size 0x%" PRIx64
+                   " is too small for 'size' option 0x" RAM_ADDR_FMT
+                   " plus 'offset' option 0x" RAM_ADDR_FMT,
+                   memory_region_name(mr), file_size, size, offset);
         return NULL;
     }