> -----Original Message-----
> From: Guoqing Jiang <guoqing.jiang@xxxxxxxxx>
> Sent: Friday, November 10, 2023 7:37 AM
> To: Bernard Metzler <BMT@xxxxxxxxxxxxxx>; linux-rdma@xxxxxxxxxxxxxxx
> Cc: jgg@xxxxxxxx; leon@xxxxxxxxxx; max7255@xxxxxxxx;
> dennis.dalessandro@xxxxxxxxxxxxxxxxxxxx; benve@xxxxxxxxx;
> vadim.fedorenko@xxxxxxxxx
> Subject: [EXTERNAL] Re: [PATCH for-next v2] RDMA/siw: Use ib_umem_get() to
> pin user pages
>
> Hi,
>
> On 11/4/23 15:56, Bernard Metzler wrote:
> > Abandon siw private code to pin user pages during user
> > memory registration, but use ib_umem_get() instead.
> > This will help maintaining the driver in case of changes
> > to the memory subsystem.
> >
> > Signed-off-by: Bernard Metzler <bmt@xxxxxxxxxxxxxx>
> > ---
> > v1 -> v2: remove RLIMIT memlock check logic, now done in ib_umem_get()
> > ---
>
> Looks good, Acked-by: Guoqing Jiang <guoqing.jiang@xxxxxxxxx>
>
> ...
>
> > + for (i = 0; num_pages > 0; i++) {
> > int nents = min_t(int, num_pages, PAGES_PER_CHUNK);
> > struct page **plist =
> > kcalloc(nents, sizeof(struct page *), GFP_KERNEL);
> >
> > if (!plist) {
> > rv = -ENOMEM;
> > - goto out_sem_up;
> > + goto err_out;
> > }
> > umem->page_chunk[i].plist = plist;
>
> One off topic question, why two dimensional list is needed for siw umem?
> Thanks in advance.
>
not needed. any list like this can be one dimensional. I wanted to keep
the page list compact in memory. I consider moving to xarray - like rxe
does - later, but wanted to move to ib_umem_get() first.
