gntdev warning fixes
Subsystems (3)
The recent gntdev fixes introduced two incorrect WARN_ON()s, which triggered immediately on my system. Patches for master and all supported stable versions follow.
PATCH 1/1 Ignore failure to unmap -1
The error paths of gntdev_mmap() can call unmap_grant_pages() even though not all of the pages have been successfully mapped. This will trigger the WARN_ON()s in __unmap_grant_pages_done(). The number of warnings can be very large; I have observed thousands of lines of warnings in the systemd journal.
Avoid this problem by only warning on unmapping failure if the handle being unmapped is not -1. The handle field of any page that was not successfully mapped will be -1, so this catches all cases wherr unmapping can legitimately fail.
--- drivers/xen/gntdev.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)
drivers/xen/gntdev.c +2 -1
@@ -416,7 +416,8 @@ static void __unmap_grant_pages_done(int result,unsigned int offset = data->unmap_ops - map->unmap_ops;for (i =0 ; i < data->count; i++) { -WARN_ON (map->unmap_ops[offset+i].status); +WARN_ON (map->unmap_ops[offset+i].status && + map->unmap_ops[offset+i].handle !=-1 );pr_debug ("unmap handle=%d st=%d\n" , map->unmap_ops[offset+i].handle, map->unmap_ops[offset+i].status);
Signature · 3 lines
Sincerely, Demi Marie Obenour (she/her/hers) Invisible Things Lab
PATCH 1/1 Ignore failure to unmap -1
In reply to Demi Marie ObenourThe error paths of gntdev_mmap() can call unmap_grant_pages() even though not all of the pages have been successfully mapped. This will trigger the WARN_ON()s in __unmap_grant_pages_done(). The number of warnings can be very large; I have observed thousands of lines of warnings in the systemd journal.
Avoid this problem by only warning on unmapping failure if the handle being unmapped is not -1. The handle field of any page that was not successfully mapped will be -1, so this catches all cases wherr unmapping can legitimately fail.
--- drivers/xen/gntdev.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)
drivers/xen/gntdev.c +2 -1
@@ -393,7 +393,8 @@ static void __unmap_grant_pages_done(int result,unsigned int offset = data->unmap_ops - map->unmap_ops;for (i =0 ; i < data->count; i++) { -WARN_ON (map->unmap_ops[offset+i].status); +WARN_ON (map->unmap_ops[offset+i].status && + map->unmap_ops[offset+i].handle !=-1 );pr_debug ("unmap handle=%d st=%d\n" , map->unmap_ops[offset+i].handle, map->unmap_ops[offset+i].status);
Signature · 3 lines
Sincerely, Demi Marie Obenour (she/her/hers) Invisible Things Lab
PATCH 1/1 Ignore failure to unmap -1
In reply to Demi Marie ObenourThe error paths of gntdev_mmap() can call unmap_grant_pages() even though not all of the pages have been successfully mapped. This will trigger the WARN_ON()s in __unmap_grant_pages_done(). The number of warnings can be very large; I have observed thousands of lines of warnings in the systemd journal.
Avoid this problem by only warning on unmapping failure if the handle being unmapped is not -1. The handle field of any page that was not successfully mapped will be -1, so this catches all cases wherr unmapping can legitimately fail.
--- drivers/xen/gntdev.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)
drivers/xen/gntdev.c +2 -1
@@ -413,7 +413,8 @@ static void __unmap_grant_pages_done(int result,unsigned int offset = data->unmap_ops - map->unmap_ops;for (i =0 ; i < data->count; i++) { -WARN_ON (map->unmap_ops[offset+i].status); +WARN_ON (map->unmap_ops[offset+i].status && + map->unmap_ops[offset+i].handle !=-1 );pr_debug ("unmap handle=%d st=%d\n" , map->unmap_ops[offset+i].handle, map->unmap_ops[offset+i].status);
Signature · 3 lines
Sincerely, Demi Marie Obenour (she/her/hers) Invisible Things Lab
PATCH 1/1 Ignore failure to unmap -1
In reply to Demi Marie ObenourThe error paths of gntdev_mmap() can call unmap_grant_pages() even though not all of the pages have been successfully mapped. This will trigger the WARN_ON()s in __unmap_grant_pages_done(). The number of warnings can be very large; I have observed thousands of lines of warnings in the systemd journal.
Avoid this problem by only warning on unmapping failure if the handle being unmapped is not -1. The handle field of any page that was not successfully mapped will be -1, so this catches all cases where unmapping can legitimately fail.
--- drivers/xen/gntdev.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)
drivers/xen/gntdev.c +2 -1
@@ -401,7 +401,8 @@ static void __unmap_grant_pages_done(int result,unsigned int offset = data->unmap_ops - map->unmap_ops;for (i =0 ; i < data->count; i++) { -WARN_ON (map->unmap_ops[offset+i].status); +WARN_ON (map->unmap_ops[offset+i].status && + map->unmap_ops[offset+i].handle !=-1 );pr_debug ("unmap handle=%d st=%d\n" , map->unmap_ops[offset+i].handle, map->unmap_ops[offset+i].status);
Signature · 3 lines
Sincerely, Demi Marie Obenour (she/her/hers) Invisible Things Lab
PATCH 1/1 Ignore failure to unmap INVALID_GRANT_HANDLE
In reply to Demi Marie ObenourThe error paths of gntdev_mmap() can call unmap_grant_pages() even though not all of the pages have been successfully mapped. This will trigger the WARN_ON()s in __unmap_grant_pages_done(). The number of warnings can be very large; I have observed thousands of lines of warnings in the systemd journal.
Avoid this problem by only warning on unmapping failure if the handle being unmapped is not INVALID_GRANT_HANDLE. The handle field of any page that was not successfully mapped will be INVALID_GRANT_HANDLE, so this catches all cases where unmapping can legitimately fail.
--- drivers/xen/gntdev.c | 8 ++++++-- 1 file changed, 6 insertions(+), 2 deletions(-)
drivers/xen/gntdev.c +6 -2
@@ -396,13 +396,17 @@ static void __unmap_grant_pages_done(int result,unsigned int offset = data->unmap_ops - map->unmap_ops;for (i =0 ; i < data->count; i++) { -WARN_ON (map->unmap_ops[offset+i].status); +WARN_ON (map->unmap_ops[offset+i].status && + map->unmap_ops[offset+i].handle != +INVALID_GRANT_HANDLE );pr_debug ("unmap handle=%d st=%d\n" , map->unmap_ops[offset+i].handle, map->unmap_ops[offset+i].status); map->unmap_ops[offset+i].handle =INVALID_GRANT_HANDLE ;if (use_ptemod) { -WARN_ON (map->kunmap_ops[offset+i].status); +WARN_ON (map->kunmap_ops[offset+i].status && + map->kunmap_ops[offset+i].handle != +INVALID_GRANT_HANDLE );pr_debug ("kunmap handle=%u st=%d\n" , map->kunmap_ops[offset+i].handle, map->kunmap_ops[offset+i].status);
Signature · 3 lines
Sincerely, Demi Marie Obenour (she/her/hers) Invisible Things Lab
PATCH 1/1 Ignore failure to unmap INVALID_GRANT_HANDLE
In reply to Demi Marie ObenourThe error paths of gntdev_mmap() can call unmap_grant_pages() even though not all of the pages have been successfully mapped. This will trigger the WARN_ON()s in __unmap_grant_pages_done(). The number of warnings can be very large; I have observed thousands of lines of warnings in the systemd journal.
Avoid this problem by only warning on unmapping failure if the handle being unmapped is not INVALID_GRANT_HANDLE. The handle field of any page that was not successfully mapped will be INVALID_GRANT_HANDLE, so this catches all cases where unmapping can legitimately fail.
--- drivers/xen/gntdev.c | 8 ++++++-- 1 file changed, 6 insertions(+), 2 deletions(-)
drivers/xen/gntdev.c +6 -2
@@ -396,13 +396,17 @@ static void __unmap_grant_pages_done(int result,unsigned int offset = data->unmap_ops - map->unmap_ops;for (i =0 ; i < data->count; i++) { -WARN_ON (map->unmap_ops[offset+i].status); +WARN_ON (map->unmap_ops[offset+i].status && + map->unmap_ops[offset+i].handle != +INVALID_GRANT_HANDLE );pr_debug ("unmap handle=%d st=%d\n" , map->unmap_ops[offset+i].handle, map->unmap_ops[offset+i].status); map->unmap_ops[offset+i].handle =INVALID_GRANT_HANDLE ;if (use_ptemod) { -WARN_ON (map->kunmap_ops[offset+i].status); +WARN_ON (map->kunmap_ops[offset+i].status && + map->kunmap_ops[offset+i].handle != +INVALID_GRANT_HANDLE );pr_debug ("kunmap handle=%u st=%d\n" , map->kunmap_ops[offset+i].handle, map->kunmap_ops[offset+i].status);
Signature · 3 lines
Sincerely, Demi Marie Obenour (she/her/hers) Invisible Things Lab
PATCH 1/1 Ignore failure to unmap INVALID_GRANT_HANDLE
In reply to Demi Marie ObenourThe error paths of gntdev_mmap() can call unmap_grant_pages() even though not all of the pages have been successfully mapped. This will trigger the WARN_ON()s in __unmap_grant_pages_done(). The number of warnings can be very large; I have observed thousands of lines of warnings in the systemd journal.
Avoid this problem by only warning on unmapping failure if the handle being unmapped is not INVALID_GRANT_HANDLE. The handle field of any page that was not successfully mapped will be INVALID_GRANT_HANDLE, so this catches all cases where unmapping can legitimately fail.
--- drivers/xen/gntdev.c | 8 ++++++-- 1 file changed, 6 insertions(+), 2 deletions(-)
drivers/xen/gntdev.c +6 -2
@@ -396,13 +396,17 @@ static void __unmap_grant_pages_done(int result,unsigned int offset = data->unmap_ops - map->unmap_ops;for (i =0 ; i < data->count; i++) { -WARN_ON (map->unmap_ops[offset+i].status); +WARN_ON (map->unmap_ops[offset+i].status && + map->unmap_ops[offset+i].handle != +INVALID_GRANT_HANDLE );pr_debug ("unmap handle=%d st=%d\n" , map->unmap_ops[offset+i].handle, map->unmap_ops[offset+i].status); map->unmap_ops[offset+i].handle =INVALID_GRANT_HANDLE ;if (use_ptemod) { -WARN_ON (map->kunmap_ops[offset+i].status); +WARN_ON (map->kunmap_ops[offset+i].status && + map->kunmap_ops[offset+i].handle != +INVALID_GRANT_HANDLE );pr_debug ("kunmap handle=%u st=%d\n" , map->kunmap_ops[offset+i].handle, map->kunmap_ops[offset+i].status);
Signature · 3 lines
Sincerely, Demi Marie Obenour (she/her/hers) Invisible Things Lab
Re: [PATCH master] Ignore failure to unmap INVALID_GRANT_HANDLE
On 08.07.22 19:37, Demi Marie Obenour wrote:
Hello Demi Marie
Please add subsystem to the commit subject, for example "xen/gntdev:".
Quoted 24 lines
The error paths of
gntdev_mmap()can callunmap_grant_pages()even though not all of the pages have been successfully mapped. This will trigger the WARN_ON()s in__unmap_grant_pages_done(). The number of warnings can be very large; I have observed thousands of lines of warnings in the systemd journal.Avoid this problem by only warning on unmapping failure if the handle being unmapped is not
INVALID_GRANT_HANDLE. The handle field of any page that was not successfully mapped will beINVALID_GRANT_HANDLE, so this catches all cases where unmapping can legitimately fail.Suggested-by Juergen Gross <[email protected]>Signed-off-by Demi Marie Obenour <[email protected]>Fixes dbe97cff7dd9 ("xen/gntdev: Avoid blocking in unmap_grant_pages()")--- drivers/xen/gntdev.c | 8 ++++++-- 1 file changed, 6 insertions(+), 2 deletions(-)drivers/xen/gntdev.c +0
-0
drivers/xen/gntdev.c static void __unmap_grant_pages_done(int result, · line 401
Quoted 7 lines
drivers/xen/gntdev.c +3
-1@@ -396,13 +396,17 @@ static void __unmap_grant_pages_done(int result,unsigned int offset = data->unmap_ops - map->unmap_ops;for (i =0 ; i < data->count; i++) { -WARN_ON (map->unmap_ops[offset+i].status); +WARN_ON (map->unmap_ops[offset+i].status && + map->unmap_ops[offset+i].handle != +INVALID_GRANT_HANDLE );
Nit: While extending the check I would clarify the first half of it:
"map->unmap_ops[offset+i].status != GNTST_okay"
drivers/xen/gntdev.c static void __unmap_grant_pages_done(int result, · line 409
Quoted 9 lines
drivers/xen/gntdev.c +3
-1pr_debug ("unmap handle=%d st=%d\n" , map->unmap_ops[offset+i].handle, map->unmap_ops[offset+i].status); map->unmap_ops[offset+i].handle =INVALID_GRANT_HANDLE ;if (use_ptemod) { -WARN_ON (map->kunmap_ops[offset+i].status); +WARN_ON (map->kunmap_ops[offset+i].status && + map->kunmap_ops[offset+i].handle != +INVALID_GRANT_HANDLE );
ditto
With updated subject:
Quoted 3 lines
drivers/xen/gntdev.c +0
-0pr_debug ("kunmap handle=%u st=%d\n" , map->kunmap_ops[offset+i].handle, map->kunmap_ops[offset+i].status);
Signature · 3 lines
Regards, Oleksandr Tyshchenko
Greg KH
maintainerQuoted 5 lines · Demi Marie Obenour
The recent gntdev fixes introduced two incorrect WARN_ON()s, which triggered immediately on my system. Patches for master and all supported stable versions follow.
<formletter>
This is not the correct way to submit patches for inclusion in the stable kernel tree. Please read:
https://www.kernel.org/doc/html/latest/process/stable-kernel-rules.html
for how to do this properly.
</formletter>