Re: [PATCH 04/10] mini-os: respect memory map when ballooning up

[Juergen Gross <jgross@suse.com>]

[-- Attachment #1.1.1: Type: text/plain, Size: 1017 bytes --]

On 12.12.21 01:26, Samuel Thibault wrote:
> Juergen Gross, le lun. 06 déc. 2021 08:23:31 +0100, a ecrit:
>> @@ -81,8 +93,11 @@ int balloon_up(unsigned long n_pages)
>>       if ( n_pages > N_BALLOON_FRAMES )
>>           n_pages = N_BALLOON_FRAMES;
>>   
>> +    start_pfn = e820_get_maxpfn(nr_mem_pages + 1) - 1;
>> +    n_pages = e820_get_max_pages(start_pfn, n_pages);
> 
> I'd say call it e820_get_max_contig_pages?

Fine with me.

> 
>> +unsigned long e820_get_max_pages(unsigned long pfn, unsigned long pages)
>> +{
>> +    int i;
>> +    unsigned long end;
>> +
>> +    for ( i = 0; i < e820_entries; i++ )
>> +    {
>> +        if ( e820_map[i].type != E820_RAM ||
>> +             (e820_map[i].addr >> PAGE_SHIFT) > pfn )
>> +            continue;
> 
> "> pfn" looks odd to me? If the start of the e820 entry is already
> beyond pfn, we'll never find any other entry. We however do want to skip
> entries that have addr+size that is below pfn.

Oh, you are right.


Juergen

[-- Attachment #1.1.2: OpenPGP public key --]
[-- Type: application/pgp-keys, Size: 3135 bytes --]

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 495 bytes --]