Skip to contentSkip to conversation

Declare PIO_CMAP/GIO_CMAP as compatbile ioctls.

PATCH In review Bill Nottingham 3 participants 5 messages Feb 10, 2009 – Feb 16, 2009 VFS

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 Declare PIO_CMAP/GIO_CMAP as compatbile ioctls.

Otherwise, these don't work when called from 32-bit userspace on 64-bit kernels.

---
 fs/compat_ioctl.c |    2 ++
 1 files changed, 2 insertions(+), 0 deletions(-)
fs/compat_ioctl.c +2 -0
index 5235c67..9ee71a0 100644
@@ -1937,6 +1937,8 @@ ULONG_IOCTL(SET_BITMAP_FILE)
 /* Big K */
 COMPATIBLE_IOCTL(PIO_FONT)
 COMPATIBLE_IOCTL(GIO_FONT)
+COMPATIBLE_IOCTL(PIO_CMAP)
+COMPATIBLE_IOCTL(GIO_CMAP)
 ULONG_IOCTL(KDSIGACCEPT)
 COMPATIBLE_IOCTL(KDGETKEYCODE)
 COMPATIBLE_IOCTL(KDSETKEYCODE)
Signature · 1 line
1.6.1.2
#1 lore raw
Quoted 20 lines · Bill Nottingham

Otherwise, these don't work when called from 32-bit userspace on 64-bit kernels.

---
 fs/compat_ioctl.c |    2 ++
 1 files changed, 2 insertions(+), 0 deletions(-)
fs/compat_ioctl.c +2 -0
index 5235c67..9ee71a0 100644
@@ -1937,6 +1937,8 @@ ULONG_IOCTL(SET_BITMAP_FILE)
 /* Big K */
 COMPATIBLE_IOCTL(PIO_FONT)
 COMPATIBLE_IOCTL(GIO_FONT)
+COMPATIBLE_IOCTL(PIO_CMAP)
+COMPATIBLE_IOCTL(GIO_CMAP)
 ULONG_IOCTL(KDSIGACCEPT)
 COMPATIBLE_IOCTL(KDGETKEYCODE)
 COMPATIBLE_IOCTL(KDSETKEYCODE)
Signature · 1 line
1.6.1.2

Alan, I am hesitant to consider this material implicitly applicable for trivial tree. Could you please give your Acked-by for it so that I can merge, or take it through your tree completely?

Thanks,

Signature · 2 lines
Jiri Kosina
SUSE Labs
#2 lore raw
Quoted 3 lines · On Wed, 11 Feb 2009 14:41:25 +0100 (CET) Jiri Kosina <[email protected]> wrote:
Quoted 1 line · Bill Nottingham

Otherwise, these don't work when called from 32-bit userspace on 64-bit kernels.

Bill, please remember to add signoffs? This should all be scripted.

Quoted 14 lines
Quoted 14 lines
---
 fs/compat_ioctl.c |    2 ++
 1 files changed, 2 insertions(+), 0 deletions(-)
fs/compat_ioctl.c +2 -0
index 5235c67..9ee71a0 100644
@@ -1937,6 +1937,8 @@ ULONG_IOCTL(SET_BITMAP_FILE)
 /* Big K */
 COMPATIBLE_IOCTL(PIO_FONT)
 COMPATIBLE_IOCTL(GIO_FONT)
+COMPATIBLE_IOCTL(PIO_CMAP)
+COMPATIBLE_IOCTL(GIO_CMAP)
Quoted 8 lines
Quoted 5 lines
fs/compat_ioctl.c +0 -0
 ULONG_IOCTL(KDSIGACCEPT)
 COMPATIBLE_IOCTL(KDGETKEYCODE)
 COMPATIBLE_IOCTL(KDSETKEYCODE)
Signature · 1 line
1.6.1.2

Alan, I am hesitant to consider this material implicitly applicable for trivial tree.

No, it isn't appropriate for the trivial tree.

Quoted 2 lines

Could you please give your Acked-by for it so that I can merge, or take it through your tree completely?

afaict this fix is needed all the way back to 2.6.tiny. I merged it with 2.6.29 intentions. Also tagged as needed in 2.6.25.x and later.

#3 lore raw

(dropping trivial@ and stable@ cc's)

Quoted 7 lines · Andrew Morton
Quoted 3 lines · On Wed, 11 Feb 2009 14:41:25 +0100 (CET) Jiri Kosina <[email protected]> wrote:
Quoted 1 line · Bill Nottingham

Otherwise, these don't work when called from 32-bit userspace on 64-bit kernels.

Bill, please remember to add signoffs? This should all be scripted.

I would have expected that that would be the default in 'git format-patch'. Will keep note of that for the future.

Quoted 5 lines
Quoted 2 lines

Could you please give your Acked-by for it so that I can merge, or take it through your tree completely?

afaict this fix is needed all the way back to 2.6.tiny. I merged it with 2.6.29 intentions. Also tagged as needed in 2.6.25.x and later.

So, is there some way to do static analysis on the kernel source to find *all* ioctls that aren't marked as compatible, and don't have specific compat handlers?

Bill

#4 lore raw
Quoted 9 lines · On Mon, 16 Feb 2009 12:23:18 -0500 Bill Nottingham <[email protected]> wrote:
Quoted 5 lines
Quoted 2 lines

Could you please give your Acked-by for it so that I can merge, or take it through your tree completely?

afaict this fix is needed all the way back to 2.6.tiny. I merged it with 2.6.29 intentions. Also tagged as needed in 2.6.25.x and later.

So, is there some way to do static analysis on the kernel source to find *all* ioctls that aren't marked as compatible, and don't have specific compat handlers?

Not that I am aware of. Julia Lawall might be able to cook up such a thing.