xen-devel.lists.xenproject.org archive mirror
 help / color / mirror / Atom feed
From: Jan Beulich <jbeulich@suse.com>
To: Julien Grall <julien@xen.org>
Cc: Stefano Stabellini <sstabellini@kernel.org>,
	Ian Jackson <iwj@xenproject.org>,
	Andrew Cooper <andrew.cooper3@citrix.com>,
	Charles Chiou <cchiou@ambarella.com>,
	"xen-devel@lists.xenproject.org" <xen-devel@lists.xenproject.org>
Subject: Re: dom0less boot two compressed kernel images out-of-memory work-around
Date: Thu, 4 Mar 2021 09:34:45 +0100	[thread overview]
Message-ID: <511603c8-40b4-5e22-7eba-7722d89895fe@suse.com> (raw)
In-Reply-To: <58aaf68f-4499-9400-6eb2-f1ad3b620c73@xen.org>

On 03.03.2021 20:36, Julien Grall wrote:
> (BCCing xen-users, CCing xen-devel + a few folks)
> 
> Hi,
> 
> Moving the discussion to xen-devel.
> 
> On 22/02/2021 05:02, Charles Chiou wrote:
>> When trying to boot two zImage using dom0less boot on ARM, we encountered this problem when xen runs gunzip on second guest:
>>
>> (XEN) ****************************************
>> (XEN) Panic on CPU 0:
>> (XEN) Out of memory
>> (XEN) ****************************************
>>
>> And worked around it with the following patch. We'd like to check to see if this is a known issue and if the work-around looks reasonable. Thank you
> 
> I haven't seen any similar report in the past.
> 
>>
>>
>> diff --git a/xen/common/gunzip.c b/xen/common/gunzip.c
>> index db4efcd34b..e5bd19ba7f 100644
>> --- a/xen/common/gunzip.c
>> +++ b/xen/common/gunzip.c
>> @@ -113,8 +113,10 @@ __init int perform_gunzip(char *output, char *image, unsigned long image_len)
>>
>>       window = (unsigned char *)output;
>>
>> +    if (!free_mem_ptr) {
>>           free_mem_ptr = (unsigned long)alloc_xenheap_pages(HEAPORDER, 0);
>>           free_mem_end_ptr = free_mem_ptr + (PAGE_SIZE << HEAPORDER);
>> +    }
>>
>>       inbuf = (unsigned char *)image;
>>       insize = image_len;
>> @@ -131,7 +133,12 @@ __init int perform_gunzip(char *output, char *image, unsigned long image_len)
>>           rc = 0;
>>       }
>>
>> +    if (free_mem_ptr) {
>>           free_xenheap_pages((void *)free_mem_ptr, HEAPORDER);
>> +        free_mem_ptr = 0;
>> +    }
>> +
>> +    bytes_out = 0;
>>
>>       return rc;
>> }
>> diff --git a/xen/common/inflate.c b/xen/common/inflate.c
>> index f99c985d61..de96002188 100644
>> --- a/xen/common/inflate.c
>> +++ b/xen/common/inflate.c
>> @@ -244,7 +244,7 @@ static void *INIT malloc(int size)
>>
>>       if (size < 0)
>>           error("Malloc error");
>> -    if (!malloc_ptr)
>> +    if ((!malloc_ptr) || (!malloc_count))
>>           malloc_ptr = free_mem_ptr;
>>
> 
> IMHO, this is a bit risky to assume that malloc_count will always be 0 
> after each gunzip.
> 
> Instead I think, it would be better if we re-initialize the allocator 
> every time.

I agree.

> How about the following (untested):
> 
> commit e1cd2d85234c8d0aa62ad32c824a5568a57be930 (HEAD -> dev)
> Author: Julien Grall <jgrall@amazon.com>
> Date:   Wed Mar 3 19:27:56 2021 +0000
> 
>      xen/gunzip: Allow perform_gunzip() to be called multiple times
> 
>      Currently perform_gunzip() can only be called once because the the
>      internal allocator is not fully re-initialized.
> 
>      This works fine if you are only booting dom0. But this will break when
>      booting multiple using the dom0less that uses compressed kernel images.
> 
>      This can be resolved by re-initializing malloc_ptr and malloc_count
>      every time perform_gunzip() is called.
> 
>      Note the latter is only re-initialized for hardening purpose as 
> there is
>      no guarantee that every malloc() are followed by free() (It should in
>      theory!).
> 
>      Take the opportunity to check the return of alloc_heap_pages() to 
> return
>      an error rather than dereferencing a NULL pointer later on failure.
> 
>      Reported-by: Charles Chiou <cchiou@ambarella.com>
>      Signed-off-by: Julien Grall <jgrall@amazon.com>
> 
>      ---
> 
>      This patch is candidate for Xen 4.15. Without this patch, it will 
> not be
>      possible to boot multiple domain using dom0less when they are using
>      compressed kernel images.

Other decompression methods are unaffected?

> --- a/xen/common/gunzip.c
> +++ b/xen/common/gunzip.c
> @@ -114,7 +114,11 @@ __init int perform_gunzip(char *output, char 
> *image, unsigned long image_len)
>       window = (unsigned char *)output;
> 
>       free_mem_ptr = (unsigned long)alloc_xenheap_pages(HEAPORDER, 0);
> +    if ( !free_mem_ptr )
> +        return -ENOMEM;
> +
>       free_mem_end_ptr = free_mem_ptr + (PAGE_SIZE << HEAPORDER);
> +    init_allocator();
> 
>       inbuf = (unsigned char *)image;
>       insize = image_len;
> diff --git a/xen/common/inflate.c b/xen/common/inflate.c
> index f99c985d6135..d8c28a3e9593 100644
> --- a/xen/common/inflate.c
> +++ b/xen/common/inflate.c
> @@ -238,6 +238,12 @@ STATIC const ush mask_bits[] = {
>   static unsigned long INITDATA malloc_ptr;
>   static int INITDATA malloc_count;
> 
> +static void init_allocator(void)

Please add INIT here. (I wouldn't mind if you used __init instead,
as there's going to be file-wide replacement after 4.15 anyway,
but of course this would render things inconsistent until then.)

Jan

> +{
> +    malloc_ptr = free_mem_ptr;
> +    malloc_count = 0;
> +}
> +
>   static void *INIT malloc(int size)
>   {
>       void *p;
> 
> Best regards,
> 



  reply	other threads:[~2021-03-04  8:35 UTC|newest]

Thread overview: 3+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
     [not found] <c1b3f92388c34788b76193b70c150521@ambarella.com>
2021-03-03 19:36 ` dom0less boot two compressed kernel images out-of-memory work-around Julien Grall
2021-03-04  8:34   ` Jan Beulich [this message]
2021-03-04 10:28     ` Julien Grall

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=511603c8-40b4-5e22-7eba-7722d89895fe@suse.com \
    --to=jbeulich@suse.com \
    --cc=andrew.cooper3@citrix.com \
    --cc=cchiou@ambarella.com \
    --cc=iwj@xenproject.org \
    --cc=julien@xen.org \
    --cc=sstabellini@kernel.org \
    --cc=xen-devel@lists.xenproject.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).