| Message ID | 1578995365-7050-1-git-send-email-magnus.karlsson@intel.com |
|---|---|
| State | Accepted |
| Delegated to: | BPF Maintainers |
| Headers | show |
| Series | [bpf-next] xsk: support allocations of large umems | expand |
On 14 Jan 2020, at 1:49, Magnus Karlsson wrote: > When registering a umem area that is sufficiently large (>1G on an > x86), kmalloc cannot be used to allocate one of the internal data > structures, as the size requested gets too large. Use kvmalloc instead > that falls back on vmalloc if the allocation is too large for kmalloc. > > Also add accounting for this structure as it is triggered by a user > space action (the XDP_UMEM_REG setsockopt) and it is by far the > largest structure of kernel allocated memory in xsk. > > Signed-off-by: Magnus Karlsson <magnus.karlsson@intel.com> > Reported-by: Ryan Goodfellow <rgoodfel@isi.edu> Acked-by: Jonathan Lemon <jonathan.lemon@gmail.com>
On Wed, Jan 15, 2020 at 11:10 AM Jonathan Lemon <jonathan.lemon@gmail.com> wrote: > > > > On 14 Jan 2020, at 1:49, Magnus Karlsson wrote: > > > When registering a umem area that is sufficiently large (>1G on an > > x86), kmalloc cannot be used to allocate one of the internal data > > structures, as the size requested gets too large. Use kvmalloc instead > > that falls back on vmalloc if the allocation is too large for kmalloc. > > > > Also add accounting for this structure as it is triggered by a user > > space action (the XDP_UMEM_REG setsockopt) and it is by far the > > largest structure of kernel allocated memory in xsk. > > > > Signed-off-by: Magnus Karlsson <magnus.karlsson@intel.com> > > Reported-by: Ryan Goodfellow <rgoodfel@isi.edu> > > Acked-by: Jonathan Lemon <jonathan.lemon@gmail.com> Applied. Thanks
diff --git a/net/xdp/xdp_umem.c b/net/xdp/xdp_umem.c index 3049af2..f93e917 100644 --- a/net/xdp/xdp_umem.c +++ b/net/xdp/xdp_umem.c @@ -249,7 +249,7 @@ static void xdp_umem_release(struct xdp_umem *umem) xdp_umem_unmap_pages(umem); xdp_umem_unpin_pages(umem); - kfree(umem->pages); + kvfree(umem->pages); umem->pages = NULL; xdp_umem_unaccount_pages(umem); @@ -409,7 +409,8 @@ static int xdp_umem_reg(struct xdp_umem *umem, struct xdp_umem_reg *mr) if (err) goto out_account; - umem->pages = kcalloc(umem->npgs, sizeof(*umem->pages), GFP_KERNEL); + umem->pages = kvcalloc(umem->npgs, sizeof(*umem->pages), + GFP_KERNEL_ACCOUNT); if (!umem->pages) { err = -ENOMEM; goto out_pin; @@ -419,7 +420,7 @@ static int xdp_umem_reg(struct xdp_umem *umem, struct xdp_umem_reg *mr) if (!err) return 0; - kfree(umem->pages); + kvfree(umem->pages); out_pin: xdp_umem_unpin_pages(umem);
When registering a umem area that is sufficiently large (>1G on an x86), kmalloc cannot be used to allocate one of the internal data structures, as the size requested gets too large. Use kvmalloc instead that falls back on vmalloc if the allocation is too large for kmalloc. Also add accounting for this structure as it is triggered by a user space action (the XDP_UMEM_REG setsockopt) and it is by far the largest structure of kernel allocated memory in xsk. Signed-off-by: Magnus Karlsson <magnus.karlsson@intel.com> Reported-by: Ryan Goodfellow <rgoodfel@isi.edu> --- net/xdp/xdp_umem.c | 7 ++++--- 1 file changed, 4 insertions(+), 3 deletions(-)