Re: [PATCH] Split segments reload in vmx_load_host_state()

[Avi Kivity <avi-atKUWr5tajBWk0Htik3J/w@public.gmane.org>]

Laurent Vivier wrote:
> Avi Kivity wrote:
>   
>> Laurent Vivier wrote:
>>     
>>> It is from the TODO and I hope no one is already working on this.
>>>
>>> 'vmx_load_host_state() bundles fs, gs, ldt, and tss reloading into one in the
>>> hope that it is infrequent. With smp guests, fs reloading is frequent and the
>>> separate reloads should be unbundled.'
>>>
>>>   
>>>       
>>>  
>>>  };
>>> @@ -353,20 +354,20 @@ static void vmx_save_host_state(struct v
>>>  	 * allow segment selectors with cpl > 0 or ti == 1.
>>>  	 */
>>>  	vmx->host_state.ldt_sel = read_ldt();
>>> -	vmx->host_state.fs_gs_ldt_reload_needed = vmx->host_state.ldt_sel;
>>> +	vmx->host_state.gs_ldt_reload_needed = vmx->host_state.ldt_sel;
>>>   
>>>       
>> Missing initialization for fs_reload_needed here.
>>
>>     
>
> It is initialized on fs_sel, should it be initialized on ldt_sel too ?
>
>   

It's only initialized to 1, but never cleared to zero.

The original code means: if ((ldt_sel != 0) || ((fs_sel  & 7) != 0) || 
((gs_sel & 7) != 0)) then a reload is needed.  In the new code, fs 
reload should depend only on ((fs_sel & 7) != 0) and ldt/gs reload 
should depend on ldt_sel and gs_sel only.

Regarding reload_tss(), it should probably be unconditional.

-- 
error compiling committee.c: too many arguments to function


-------------------------------------------------------------------------
This SF.net email is sponsored by: Splunk Inc.
Still grepping through log files to find problems?  Stop.
Now Search log events and configuration files using AJAX and a browser.
Download your FREE copy of Splunk now >>  http://get.splunk.com/