Re: [patch 06/11] Text Edit Lock - Alternative code for x86

[Mathieu Desnoyers <mathieu.desnoyers@polymtl.ca>]

* pageexec@freemail.hu (pageexec@freemail.hu) wrote:
> On 13 Nov 2007 at 13:46, Mathieu Desnoyers wrote:
> 
> > +void *text_poke_early(void *addr, const void *opcode, size_t len)
> > +{
> > +	memcpy(addr, opcode, len);
> > +	text_sync(addr, len);
> > +	return addr;
> > +}
> 
> why do you need this function (vs. using text_poke throughout)?
> 

Because it's not safe to use read_cr0() in paravirtualization
before the alternatives are set.

> > +#define kernel_wp_save(cr0)					\
> > +	do {							\
> > +		preempt_disable();				\
> > +		cr0 = read_cr0();				\
> > +		if (cpu_data(smp_processor_id()).wp_works_ok)	\
> 
> why do you need this test? if cr0.wp is ineffective, then it doesn't
> matter whether it's on or off (in fact, at least the intel manual
> says that 386s would not even let you change its value, they'll
> silently ignore attempts of setting the wp bit).
> 

Ok.. then this test could go away then. I prefered to use a conservative
approach. Will fix. Thanks for the hint.

Mathieu

> > +			write_cr0(cr0 & ~X86_CR0_WP);		\
> > +	} while (0)
> > +
> > +#define kernel_wp_restore(cr0)					\
> > +	do {							\
> > +		if (cpu_data(smp_processor_id()).wp_works_ok)	\
> 
> ditto...
> 
> > +			write_cr0(cr0);				\
> > +		preempt_enable();				\
> > +	} while (0)
> >  
> >  #endif /* _I386_ALTERNATIVE_H */
> 
> 

-- 
Mathieu Desnoyers
Computer Engineering Ph.D. Student, Ecole Polytechnique de Montreal
OpenPGP key fingerprint: 8CD5 52C3 8E3C 4140 715F  BA06 3F25 A8FE 3BAE 9A68