On Fri, Dec 7, 2018 at 12:23 AM Jason Gunthorpe <jgg@xxxxxxxxxxxx> wrote:
>
> On Thu, Dec 06, 2018 at 10:49:43AM -0800, Selvin Xavier wrote:
> > From: Somnath Kotur <somnath.kotur@xxxxxxxxxxxx>
> >
> > Currently bnxt_re driver is not utilizing the full limits of our h/w
> > to register single MR. Increase the maximum MR size to 256G.
> > The patch includes the following changes.
> >
> > - Use 'vmalloc' instead of 'kmalloc' while trying to allocate a
> > temporary place holder for all the pointers to user memory pages.
> > - Determine how many logical PDEs are required to store the pointers
> > to user memory pages and allocate contigous pages of sufficient
> > size to store all of them.
>
> You should really work with Shiraz on his patches in this area. It
> might able able to eliminate the expensive temporary memory.
>
> > - if (!umem->hugetlb && length > BNXT_RE_MAX_MR_SIZE_LOW) {
> > - dev_err(rdev_to_dev(rdev), "Requested MR Sz:%llu Max sup:%llu",
> > - length, (u64)BNXT_RE_MAX_MR_SIZE_LOW);
> > - rc = -EINVAL;
> > - goto fail;
> > - }
> > if (umem->hugetlb && length > BNXT_RE_PAGE_SIZE_2M) {
> > page_shift = BNXT_RE_PAGE_SHIFT_2M;
> > dev_warn(rdev_to_dev(rdev), "umem hugetlb set page_size %x",
> > 1 << page_shift);
> > }
>
> Use of umem->hugetlb needs to be removed from this driver, it is the
> last (wrong) user left.
Sure. I will post a separate patch to remove this. Is that okay?
>
> Jason
