Re: [PATCH RESEND v2] x86: introduce ioremap_wc()

[Jan Beulich <jbeulich@xxxxxxxx>]

On 18.02.2022 10:09, Roger Pau Monné wrote:
> On Thu, Feb 17, 2022 at 04:57:41PM +0100, Jan Beulich wrote:
>> On 17.02.2022 16:50, Roger Pau Monné wrote:
>>> On Thu, Feb 17, 2022 at 04:02:39PM +0100, Jan Beulich wrote:
>>>> On 17.02.2022 15:47, Roger Pau Monné wrote:
>>>>> On Thu, Feb 17, 2022 at 12:01:08PM +0100, Jan Beulich wrote:
>>>>>> --- a/xen/drivers/video/vga.c
>>>>>> +++ b/xen/drivers/video/vga.c
>>>>>> @@ -79,7 +79,7 @@ void __init video_init(void)
>>>>>>      {
>>>>>>      case XEN_VGATYPE_TEXT_MODE_3:
>>>>>>          if ( page_is_ram_type(paddr_to_pfn(0xB8000), 
>>>>>> RAM_TYPE_CONVENTIONAL) ||
>>>>>> -             ((video = ioremap(0xB8000, 0x8000)) == NULL) )
>>>>>> +             ((video = ioremap_wc(0xB8000, 0x8000)) == NULL) )
>>>>>>              return;
>>>>>>          outw(0x200a, 0x3d4); /* disable cursor */
>>>>>>          columns = vga_console_info.u.text_mode_3.columns;
>>>>>> @@ -164,7 +164,11 @@ void __init video_endboot(void)
>>>>>>      {
>>>>>>      case XEN_VGATYPE_TEXT_MODE_3:
>>>>>>          if ( !vgacon_keep )
>>>>>> +        {
>>>>>>              memset(video, 0, columns * lines * 2);
>>>>>> +            iounmap(video);
>>>>>> +            video = ZERO_BLOCK_PTR;
>>>>>> +        }
>>>>>>          break;
>>>>>>      case XEN_VGATYPE_VESA_LFB:
>>>>>>      case XEN_VGATYPE_EFI_LFB:
>>>>>
>>>>> I think in vesa_endboot you also need to iounmap the framebuffer
>>>>> iomem?
>>>>
>>>> Again part of the full series. I guess I was a little inconsistent
>>>> with leaving the VGA unmap in here, but breaking out the VESA part.
>>>> It's been a long time, but I guess I did so because the VESA part
>>>> needs to touch two files.
>>>
>>> I think you are hesitant to include the chunks for the above items? (or
>>> maybe I'm not properly accounting for their complexity).
>>
>> There's no complexity, it's really just that the zapping of the pointer
>> needs to be done in a different place from where the unmap is. See below.
>>
>> Jan
>>
>> video/vesa: unmap frame buffer when relinquishing console
>>
>> There's no point in keeping the VA space occupied when no further output
>> will occur.
>>
>> Signed-off-by: Jan Beulich <jbeulich@xxxxxxxx>
> 
> Reviewed-by: Roger Pau Monné <roger.pau@xxxxxxxxxx>
> 
> For both patches, the one inline here and "x86: introduce
> ioremap_wc()".

Thanks. Actually, while looking back at the original thread, to re-check
what pending objections there might have been, I did find the reason for
the split: In the patch here I would have introduced another leak, while
the other patch fixes an existing one.

> While at it, I think you should also push "video/vesa: drop
> "vesa-mtrr" command line option".

Yes, that one's merely dependent on the one here.

Jan