Re: [PATCH 1/2] x86: Add Isolated Memory Regions for Quark X1000

["Bryan O'Donoghue" <pure.logic@nexus-software.ie>]

On 08/01/15 00:04, Ong, Boon Leong wrote:

Hi Boon Leong - skipping the simple stuff.

>> +/**
>> + * imr_del_range - Delete an Isolated Memory Region
>> + * @reg: IMR index to remove
>> + * @base: Physical base address of region aligned to 4k
>> + * @size: Physical size of region in bytes
>> + * @return:	-EINVAL on invalid range or out or range id
>> + *		-ENODEV if reg is valid but no IMR exists or is locked
>> + *		0 on success
>> + */
>> +int imr_del(int reg, unsigned long base, unsigned long size);
>
> How about just offer imr delete based index-only as returned by imr_add()?
> We just need to check if that IMR is locked. If locked, then bail out.
> Else, we will zero-out IMR register for that index to remove it.

Hmm.

The MTRR API this is based on allows you to specific an address range 
and I think that makes sense for an IMR API too because - say you want 
to tear down the IMR around the kernel .text area - but you don't know 
which IMR it is.

You'd just do

unsigned long base  = virt_to_phys(&_text);
unsigned long size = virt_to_phys(&_sinittext) - base - IMR_ALIGN;

imr_del(-1, base, size);

Rather than having to know which specific index to kill. Also later 
silicon may have more - or less IMR indices - so deleting based on an 
address range is a valuable feature I think.

> if (!x86_match_cpu(soc_imr_ids) || !iosf_mbi_available()) {
> 	pr_info("IMR init failed due to IOSF_MBI not available or SoC is not Quark.\n");
> 	return -ENODEV;
> } else {
> 	imr_dev.num = QUARK_X1000_IMR_NUM;
> 	imr_dev.reg_base = QUARK_X1000_IMR_REGBASE;
> }

That works for me.

--
BOD