Messages in this thread | | | Date | Thu, 27 Jul 2023 13:40:21 +0200 | Subject | Re: [Linaro-mm-sig] Re: [RFC PATCH 00/10] Device Memory TCP | From | Christian König <> |
| |
Am 20.07.23 um 01:24 schrieb Jason Gunthorpe: > On Wed, Jul 19, 2023 at 10:57:11AM -0700, Stephen Hemminger wrote: > >> Naive idea. >> Would it be possible for process to use mmap() on the GPU memory and then >> do zero copy TCP receive some how? Or is this what is being proposed. > It could be possible, but currently there is no API to recover the > underlying dmabuf from the VMA backing the mmap.
Sorry for being a bit late, have been on vacation.
Well actually this was discussed before to work around problems with Windows applications through wine/proton.
Not 100% sure what the outcome of that was, but if I'm not completely mistaken getting the fd behind a VMA should be possible.
It might just not be the DMA-buf fd, because we use mmap() re-routing to be able to work around problems with the reverse tracking of mappings.
Christian.
> > Also you can't just take arbitary struct pages from any old VMA and > make them "netmem" > > Jason > _______________________________________________ > Linaro-mm-sig mailing list -- linaro-mm-sig@lists.linaro.org > To unsubscribe send an email to linaro-mm-sig-leave@lists.linaro.org
| |