| Message ID | 20200824183641.632126-1-jhubbard@nvidia.com (mailing list archive) |
|---|---|
| State | Superseded |
| Headers | show |
| Series | [v2] tee: convert convert get_user_pages() --> pin_user_pages() | expand |
On 8/24/20 11:36 AM, John Hubbard wrote: > This code was using get_user_pages*(), in a "Case 2" scenario > (DMA/RDMA), using the categorization from [1]. That means that it's > time to convert the get_user_pages*() + put_page() calls to > pin_user_pages*() + unpin_user_pages() calls. > > There is some helpful background in [2]: basically, this is a small > part of fixing a long-standing disconnect between pinning pages, and > file systems' use of those pages. > > [1] Documentation/core-api/pin_user_pages.rst > > [2] "Explicit pinning of user-space pages": > https://lwn.net/Articles/807108/ > > Cc: Jens Wiklander <jens.wiklander@linaro.org> > Cc: Sumit Semwal <sumit.semwal@linaro.org> > Cc: tee-dev@lists.linaro.org > Cc: linux-media@vger.kernel.org > Cc: dri-devel@lists.freedesktop.org > Cc: linaro-mm-sig@lists.linaro.org > Signed-off-by: John Hubbard <jhubbard@nvidia.com> > --- > > OK, this should be indentical to v1 [1], but now rebased against > Linux 5.9-rc2. > ...ohhh, wait, I should have read the earlier message from Jens more carefully: "The conflict isn't trivial, I guess we need to handle the different types of pages differently when releasing them." So it's not good to have a logically identical patch. argghhh. Let me see how hard it is to track these memory types separately and handle the release accordingly, just a sec. Sorry about the false move here. thanks,
diff --git a/drivers/tee/tee_shm.c b/drivers/tee/tee_shm.c index 827ac3d0fea9..3c29e6c3ebe8 100644 --- a/drivers/tee/tee_shm.c +++ b/drivers/tee/tee_shm.c @@ -32,16 +32,13 @@ static void tee_shm_release(struct tee_shm *shm) poolm->ops->free(poolm, shm); } else if (shm->flags & TEE_SHM_REGISTER) { - size_t n; int rc = teedev->desc->ops->shm_unregister(shm->ctx, shm); if (rc) dev_err(teedev->dev.parent, "unregister shm %p failed: %d", shm, rc); - for (n = 0; n < shm->num_pages; n++) - put_page(shm->pages[n]); - + unpin_user_pages(shm->pages, shm->num_pages); kfree(shm->pages); } @@ -228,7 +225,7 @@ struct tee_shm *tee_shm_register(struct tee_context *ctx, unsigned long addr, } if (flags & TEE_SHM_USER_MAPPED) { - rc = get_user_pages_fast(start, num_pages, FOLL_WRITE, + rc = pin_user_pages_fast(start, num_pages, FOLL_WRITE, shm->pages); } else { struct kvec *kiov; @@ -292,16 +289,13 @@ struct tee_shm *tee_shm_register(struct tee_context *ctx, unsigned long addr, return shm; err: if (shm) { - size_t n; - if (shm->id >= 0) { mutex_lock(&teedev->mutex); idr_remove(&teedev->idr, shm->id); mutex_unlock(&teedev->mutex); } if (shm->pages) { - for (n = 0; n < shm->num_pages; n++) - put_page(shm->pages[n]); + unpin_user_pages(shm->pages, shm->num_pages); kfree(shm->pages); } }
This code was using get_user_pages*(), in a "Case 2" scenario (DMA/RDMA), using the categorization from [1]. That means that it's time to convert the get_user_pages*() + put_page() calls to pin_user_pages*() + unpin_user_pages() calls. There is some helpful background in [2]: basically, this is a small part of fixing a long-standing disconnect between pinning pages, and file systems' use of those pages. [1] Documentation/core-api/pin_user_pages.rst [2] "Explicit pinning of user-space pages": https://lwn.net/Articles/807108/ Cc: Jens Wiklander <jens.wiklander@linaro.org> Cc: Sumit Semwal <sumit.semwal@linaro.org> Cc: tee-dev@lists.linaro.org Cc: linux-media@vger.kernel.org Cc: dri-devel@lists.freedesktop.org Cc: linaro-mm-sig@lists.linaro.org Signed-off-by: John Hubbard <jhubbard@nvidia.com> --- OK, this should be indentical to v1 [1], but now rebased against Linux 5.9-rc2. As before, I've compile-tested it again with a cross compiler, but that's the only testing I'm set up for with CONFIG_TEE. [1] https://lore.kernel.org/r/20200519051850.2845561-1-jhubbard@nvidia.com thanks, John Hubbard NVIDIA drivers/tee/tee_shm.c | 12 +++--------- 1 file changed, 3 insertions(+), 9 deletions(-)