xen/gntdev.c: Mark pages as dirty
Subsystems (2)
PATCH 1/2 xen/gntdev.c: Mark pages as dirty
There seems to be a bug in the original code when gntdev_get_page() is called with writeable=true then the page needs to be marked dirty before being put.
To address this, a bool writeable is added in gnt_dev_copy_batch, set it in gntdev_grant_copy_seg() (and drop writeable argument to gntdev_get_page()) and then, based on batch->writeable, use set_page_dirty_lock().
--- drivers/xen/gntdev.c | 17 ++++++++++++----- 1 file changed, 12 insertions(+), 5 deletions(-)
drivers/xen/gntdev.c +12 -5
@@ -720,17 +720,18 @@ struct gntdev_copy_batch {s16 __user *status[GNTDEV_COPY_BATCH ];unsigned int nr_ops;unsigned int nr_pages; +bool writeable; };static int gntdev_get_page (struct gntdev_copy_batch *batch,void __user *virt, -bool writeable,unsigned long *gfn) +unsigned long *gfn) {unsigned long addr = (unsigned long )virt;struct page *page;unsigned long xen_pfn;int ret; - ret =get_user_pages_fast (addr,1 , writeable ?FOLL_WRITE :0 , &page); + ret =get_user_pages_fast (addr,1 , batch->writeable ?FOLL_WRITE :0 , &page);if (ret <0 )return ret;
@@ -746,9 +747,13 @@ static void gntdev_put_pages(struct gntdev_copy_batch *batch) {unsigned int i; -for (i =0 ; i < batch->nr_pages; i++) +for (i =0 ; i < batch->nr_pages; i++) { +if (batch->writeable && !PageDirty (batch->pages[i])) +set_page_dirty_lock (batch->pages[i]);put_page (batch->pages[i]); + } batch->nr_pages =0 ; + batch->writeable = false; }static int gntdev_copy (struct gntdev_copy_batch *batch)
@@ -837,8 +842,9 @@ static int gntdev_grant_copy_seg(struct gntdev_copy_batch *batch, virt = seg->source.virt + copied; off = (unsigned long )virt & ~XEN_PAGE_MASK ; len =min (len, (size_t )XEN_PAGE_SIZE - off); + batch->writeable = false; - ret =gntdev_get_page (batch, virt, false, &gfn); + ret =gntdev_get_page (batch, virt, &gfn);if (ret <0 )return ret;
@@ -856,8 +862,9 @@ static int gntdev_grant_copy_seg(struct gntdev_copy_batch *batch, virt = seg->dest.virt + copied; off = (unsigned long )virt & ~XEN_PAGE_MASK ; len =min (len, (size_t )XEN_PAGE_SIZE - off); + batch->writeable = true; - ret =gntdev_get_page (batch, virt, true, &gfn); + ret =gntdev_get_page (batch, virt, &gfn);if (ret <0 )return ret;
Signature · 1 line
1.9.1
PATCH 2/2 xen/gntdev.c: Convert get_user_pages*() to pin_user_pages*()
In 2019, we introduced pin_user_pages*() and now we are converting get_user_pages*() to the new API as appropriate. [1] & [2] could be referred for more information. This is case 5 as per document [1].
[1] Documentation/core-api/pin_user_pages.rst
[2] "Explicit pinning of user-space pages":
https://lwn.net/Articles/807108/
--- drivers/xen/gntdev.c | 10 ++-------- 1 file changed, 2 insertions(+), 8 deletions(-)
drivers/xen/gntdev.c +2 -8
@@ -731,7 +731,7 @@ static int gntdev_get_page(struct gntdev_copy_batch *batch, void __user *virt,unsigned long xen_pfn;int ret; - ret =get_user_pages_fast (addr,1 , batch->writeable ?FOLL_WRITE :0 , &page); + ret =pin_user_pages_fast (addr,1 , batch->writeable ?FOLL_WRITE :0 , &page);if (ret <0 )return ret;
@@ -745,13 +745,7 @@ static int gntdev_get_page(struct gntdev_copy_batch *batch, void __user *virt,static void gntdev_put_pages (struct gntdev_copy_batch *batch) { -unsigned int i; - -for (i =0 ; i < batch->nr_pages; i++) { -if (batch->writeable && !PageDirty (batch->pages[i])) -set_page_dirty_lock (batch->pages[i]); -put_page (batch->pages[i]); - } +unpin_user_pages_dirty_lock (batch->pages, batch->nr_pages, batch->writeable); batch->nr_pages =0 ; batch->writeable = false; }
Signature · 1 line
1.9.1
Re: [PATCH 2/2] xen/gntdev.c: Convert get_user_pages*() to pin_user_pages*()
Quoted 6 lines · Souptick Joarder
In 2019, we introduced pin_user_pages*() and now we are converting get_user_pages*() to the new API as appropriate. [1] & [2] could be referred for more information. This is case 5 as per document [1].
[1]
Documentation/core-api/pin_user_pages.rst
Quoted 8 lines
[2] "Explicit pinning of user-space pages":
https://lwn.net/Articles/807108/Signed-off-by Souptick Joarder <[email protected]>Cc John Hubbard <[email protected]>Cc Boris Ostrovsky <[email protected]>Cc Juergen Gross <[email protected]>Cc David Vrabel <[email protected]>
Re: [PATCH 2/2] xen/gntdev.c: Convert get_user_pages*() to pin_user_pages*()
Quoted 12 lines · On Fri, Sep 11, 2020 at 8:12 PM <[email protected]> wrote:
Quoted 9 lines · Souptick Joarder
In 2019, we introduced pin_user_pages*() and now we are converting get_user_pages*() to the new API as appropriate. [1] & [2] could be referred for more information. This is case 5 as per document [1].
[1]
Documentation/core-api/pin_user_pages.rst[2] "Explicit pinning of user-space pages":
https://lwn.net/Articles/807108/
Quoted 8 lines
Quoted 5 lines
Signed-off-by Souptick Joarder <[email protected]>Cc John Hubbard <[email protected]>Cc Boris Ostrovsky <[email protected]>Cc Juergen Gross <[email protected]>Cc David Vrabel <[email protected]>Reviewed-by Boris Ostrovsky <[email protected]>
Are these 2 patches queued for 5.10-rc1 ?
Quoted 2 lines
Re: [PATCH 2/2] xen/gntdev.c: Convert get_user_pages*() to pin_user_pages*()
Quoted 12 lines · Souptick Joarder
Quoted 11 lines · On Fri, Sep 11, 2020 at 8:12 PM <[email protected]> wrote:
Quoted 9 lines · Souptick Joarder
In 2019, we introduced pin_user_pages*() and now we are converting get_user_pages*() to the new API as appropriate. [1] & [2] could be referred for more information. This is case 5 as per document [1].
[1]
Documentation/core-api/pin_user_pages.rst[2] "Explicit pinning of user-space pages":
https://lwn.net/Articles/807108/
Quoted 8 lines
Quoted 7 lines
Quoted 5 lines
Signed-off-by Souptick Joarder <[email protected]>Cc John Hubbard <[email protected]>Cc Boris Ostrovsky <[email protected]>Cc Juergen Gross <[email protected]>Cc David Vrabel <[email protected]>Reviewed-by Boris Ostrovsky <[email protected]>Are these 2 patches queued for 5.10-rc1 ?
Yes, I am preparing the branch. (BTW, your second patch appears to have been either manually edited or not generated on top of the first patch. Please don't do this next time)
-boris
Re: [PATCH 2/2] xen/gntdev.c: Convert get_user_pages*() to pin_user_pages*()
Quoted 21 lines · On Tue, Sep 29, 2020 at 6:00 PM <[email protected]> wrote:
Quoted 18 lines · Souptick Joarder
Quoted 17 lines · On Fri, Sep 11, 2020 at 8:12 PM <[email protected]> wrote:
Quoted 14 lines · Souptick Joarder
In 2019, we introduced pin_user_pages*() and now we are converting get_user_pages*() to the new API as appropriate. [1] & [2] could be referred for more information. This is case 5 as per document [1].
[1]
Documentation/core-api/pin_user_pages.rst[2] "Explicit pinning of user-space pages":
https://lwn.net/Articles/807108/Signed-off-by Souptick Joarder <[email protected]>Cc John Hubbard <[email protected]>Cc Boris Ostrovsky <[email protected]>Cc Juergen Gross <[email protected]>Cc David Vrabel <[email protected]>
Quoted 5 lines
Quoted 2 lines
Quoted 1 line
Reviewed-by Boris Ostrovsky <[email protected]>Are these 2 patches queued for 5.10-rc1 ?
Yes, I am preparing the branch. (BTW, your second patch appears to have been either manually edited or not generated on top of the first patch. Please don't do this next time)
I created it on top of the first one and didn't edit manually. I was able to apply it in my local repository.
What was the error ?
Re: [PATCH 2/2] xen/gntdev.c: Convert get_user_pages*() to pin_user_pages*()
Quoted 23 lines · Souptick Joarder
Quoted 22 lines · On Tue, Sep 29, 2020 at 6:00 PM <[email protected]> wrote:
Quoted 19 lines · Souptick Joarder
Quoted 18 lines · On Fri, Sep 11, 2020 at 8:12 PM <[email protected]> wrote:
Quoted 6 lines · Souptick Joarder
In 2019, we introduced pin_user_pages*() and now we are converting get_user_pages*() to the new API as appropriate. [1] & [2] could be referred for more information. This is case 5 as per document [1].
[1]
Documentation/core-api/pin_user_pages.rstQuoted 8 lines
[2] "Explicit pinning of user-space pages":
https://lwn.net/Articles/807108/Signed-off-by Souptick Joarder <[email protected]>Cc John Hubbard <[email protected]>Cc Boris Ostrovsky <[email protected]>Cc Juergen Gross <[email protected]>Cc David Vrabel <[email protected]>Reviewed-by Boris Ostrovsky <[email protected]>
Quoted 9 lines
Quoted 4 lines
Quoted 1 line
Are these 2 patches queued for 5.10-rc1 ?
Yes, I am preparing the branch. (BTW, your second patch appears to have been either manually edited or not generated on top of the first patch. Please don't do this next time)
I created it on top of the first one and didn't edit manually. I was able to apply it in my local repository.
What was the error ?
Patch 1:
+ if (batch->writeable && !PageDirty(batch->pages[i]))
Patch 2:
- if(batch->writeable && !PageDirty(batch->pages[i]))
This doesn't look to me like usual whitespace damage in-flight. Anyway, this has been applied to for-linus-5.10
-boris
Quoted 9 lines · Souptick Joarder
There seems to be a bug in the original code when
gntdev_get_page()is called with writeable=true then the page needs to be marked dirty before being put.To address this, a bool writeable is added in
gnt_dev_copy_batch, set it ingntdev_grant_copy_seg()(and dropwriteableargument togntdev_get_page()) and then, based onbatch->writeable, useset_page_dirty_lock().
Quoted 7 lines
Fixes a4cdb556cae0 (xen/gntdev: add ioctl for grant copy)Suggested-by Boris Ostrovsky <[email protected]>Signed-off-by Souptick Joarder <[email protected]>Cc John Hubbard <[email protected]>Cc Boris Ostrovsky <[email protected]>Cc Juergen Gross <[email protected]>Cc David Vrabel <[email protected]>
(can be added at commit time)