On Wed, Jun 7, 2017 at 2:01 PM, Vitaly Kuznetsov <vkuznets@xxxxxxxxxx> wrote:
> Hyper-V hosts may support more than 64 vCPUs, we need to use
> HVCALL_FLUSH_VIRTUAL_ADDRESS_SPACE_EX/LIST_EX hypercalls in this
> case.
> +/* HvFlushVirtualAddressSpaceEx, HvFlushVirtualAddressListEx hypercalls */
> +struct hv_flush_pcpu_ex {
> + u64 address_space;
> + u64 flags;
> + struct {
> + u64 format;
> + u64 valid_bank_mask;
> + u64 bank_contents[];
> + } hv_vp_set;
> + u64 gva_list[];
> +};
> +static struct hv_flush_pcpu_ex __percpu *pcpu_flush_ex;
> - flush->address_space = virt_to_phys(mm->pgd);
> + flush->address_space = (u64)virt_to_phys(mm->pgd);
I think this casting is redundant. Your variable fits phys_addr_t
always. Like you do below w/o explicit casting.
> +static void hyperv_flush_tlb_others_ex(const struct cpumask *cpus,
> + struct mm_struct *mm,
> + unsigned long start,
> + unsigned long end)
> +{
> + int nr_bank = 0, max_gvas, gva_n;
> + struct hv_flush_pcpu_ex *flush;
> + u64 status = U64_MAX;
In one of the previous patches you used (u64)ULLONG_MAX.
I recommend to use _there_ same as here, i.e. = U64_MAX;
> + if (mm) {
> + flush->address_space = virt_to_phys(mm->pgd);
No explicit casting is okay here.
> + flush->flags = 0;
> + } else {
> + flush->address_space = 0;
> + flush->flags = HV_FLUSH_ALL_VIRTUAL_ADDRESS_SPACES;
> + }
> + /*
> + * We can flush not more than max_gvas with one hypercall. Flush the
> + * whole address space if we were asked to do more.
> + */
> + max_gvas = (PAGE_SIZE - sizeof(*flush) -
> + nr_bank * sizeof(flush->hv_vp_set.bank_contents[0])) /
> + sizeof(flush->gva_list[0]);
Is it possible to re-indent like
max_gvas =
(PAGE_SIZE - sizeof(*flush) - nr_bank *
sizeof(flush->hv_vp_set.bank_contents[0])) /
sizeof(flush->gva_list[0]);
for easier understanding the calculus?
--
With Best Regards,
Andy Shevchenko
_______________________________________________
devel mailing list
devel@xxxxxxxxxxxxxxxxxxxxxx
http://driverdev.linuxdriverproject.org/mailman/listinfo/driverdev-devel
