Re: [PATCH 2/2] habanalabs: allow user to set allocation page size

From: Oded Gabbay
Date: Mon Feb 14 2022 - 08:04:03 EST


On Mon, Feb 14, 2022 at 3:02 PM Oded Gabbay <ogabbay@xxxxxxxxxx> wrote:
>
> On Mon, Feb 14, 2022 at 12:48 PM Hillf Danton <hdanton@xxxxxxxx> wrote:
> >
> > On Mon, 14 Feb 2022 10:12:10 +0200 Ohad Sharabi wrote:
> > >
> > > struct hl_mem_in {
> > > union {
> > > - /* HL_MEM_OP_ALLOC- allocate device memory */
> > > + /**
> > > + * structure for device memory allocation (used with the HL_MEM_OP_ALLOC op)
> > > + * @mem_size: memory size to allocate
> > > + * @page_size: page size to use on allocation. when the value is 0 the default page
> > > + * size will be taken.
> > > + */
> >
> > Nit, the 80-column-char format is prfered.
> Maybe it is preferred, but afaik it is now allowed to use 100-chars,
> and checkpatch no longer shouts about
> it, so I don't see the harm.
> >
> > Wonder how to get the available page sizes in userspace?
> We will add an opcode to our INFO ioctl to return the available page sizes.
> The user will need to query this at the beginning of the application run.
This will be in a separate patch.
Oded

>
>
> >
> > > struct {
> > > - /* Size to alloc */
> > > __u64 mem_size;
> > > + __u64 page_size;
> > > } alloc;
> > >
> > > - /* HL_MEM_OP_FREE - free device memory */
> > > + /**
> > > + * structure for free-ing device memory (used with the HL_MEM_OP_FREE op)
> > > + * @handle: handle returned from HL_MEM_OP_ALLOC
> >
> > Nit, add a seperate cleanup patch for changes with no link to page size.
> Yes, correct, I will split this up.
>
> Thanks,
> Oded
> >
> > > + */
> > > struct {
> > > - /* Handle returned from HL_MEM_OP_ALLOC */
> > > __u64 handle;
> > > } free;
> > >
> > > - /* HL_MEM_OP_MAP - map device memory */
> > > + /**
> > > + * structure for mapping device memory (used with the HL_MEM_OP_MAP op)
> > > + * @hint_addr: requested virtual address of mapped memory.
> > > + * the driver will try to map the requested region to this hint
> > > + * address, as long as the address is valid and not already mapped.
> > > + * the user should check the returned address of the IOCTL to make
> > > + * sure he got the hint address.
> > > + * passing 0 here means that the driver will choose the address itself.
> > > + * @handle: handle returned from HL_MEM_OP_ALLOC.
> > > + */
> > > struct {
> > > - /*
> > > - * Requested virtual address of mapped memory.
> > > - * The driver will try to map the requested region to
> > > - * this hint address, as long as the address is valid
> > > - * and not already mapped. The user should check the
> > > - * returned address of the IOCTL to make sure he got
> > > - * the hint address. Passing 0 here means that the
> > > - * driver will choose the address itself.
> > > - */
> > > __u64 hint_addr;
> > > - /* Handle returned from HL_MEM_OP_ALLOC */
> > > __u64 handle;
> > > } map_device;