On Thu, Jul 04, 2013 at 06:39:03PM +0200, Stefan Fritsch wrote:
> diff --git sys/netinet/ip_output.c sys/netinet/ip_output.c
> index b59accf..43a0551 100644
> --- sys/netinet/ip_output.c
> +++ sys/netinet/ip_output.c
> @@ -267,7 +267,7 @@ reroute:
> if (mtag != NULL) {
> #ifdef DIAGNOSTIC
>
> Date: Thu, 4 Jul 2013 18:42:50 +0200 (CEST)
> From: Stefan Fritsch
>
> On Wed, 3 Jul 2013, Mark Kettenis wrote:
> > > diff --git sys/arch/i386/i386/esm.c sys/arch/i386/i386/esm.c
> > > index c90b2c4..3dff69e 100644
> > > --- sys/arch/i386/i386/esm.c
> > > +++ sys/arch/i386/i386/esm.c
> > > @@ -
> Date: Thu, 4 Jul 2013 18:41:30 +0200 (CEST)
> From: Stefan Fritsch
>
> On Wed, 3 Jul 2013, Mark Kettenis wrote:
> > > diff --git sys/arch/i386/i386/db_interface.c
> > > sys/arch/i386/i386/db_interface.c
> > > index 85c1ff5..c75fd89 100644
> > > --- sys/arch/i386/i386/db_interface.c
> > > +++ s
On Jul 4, 2013, at 6:43 PM, Stefan Fritsch wrote:
> fix: %x instead of %p for int
>
> ---
> sys/dev/pci/musycc_obsd.c |2 +-
> 1 files changed, 1 insertions(+), 1 deletions(-)
>
> diff --git sys/dev/pci/musycc_obsd.c sys/dev/pci/musycc_obsd.c
> index 25a58d8..0844136 100644
> --- sys/dev/pci
On Wednesday 03 July 2013, Stefan Fritsch wrote:
> I haven't done any signedness fixes so far, only the size fixes.
> But I will look over the patches again and try to fix the
> signedness issues in the touched code lines, too. There are quite
> a few I guess.
For people prefering to review by sou
%hu/%hd for uint16_t, %u/%d/%x for uint32_t
despite the name, ntohl returns uint32_t
also fix some %d into %u
---
sys/arch/i386/i386/bios.c|4 ++--
sys/arch/i386/i386/ioapic.c |2 +-
sys/arch/i386/pci/pcibios.c |4 ++--
sys/kern/vfs_cluster.c |2 +-
sys/msdosfs/msdosf
paddr_t is long
---
sys/arch/i386/i386/machdep.c |4 ++--
sys/arch/i386/pci/pci_addr_fixup.c |2 +-
2 files changed, 3 insertions(+), 3 deletions(-)
diff --git sys/arch/i386/i386/machdep.c sys/arch/i386/i386/machdep.c
index 7924386..1a26e7d 100644
--- sys/arch/i386/i386/machdep.c
fix: pointer to long
---
sys/arch/i386/pci/pci_addr_fixup.c |2 +-
1 files changed, 1 insertions(+), 1 deletions(-)
diff --git sys/arch/i386/pci/pci_addr_fixup.c sys/arch/i386/pci/pci_addr_fixup.c
index 59880eb..40f7918 100644
--- sys/arch/i386/pci/pci_addr_fixup.c
+++ sys/arch/i386/pci/pci
fix: %x instead of %p for int
---
sys/dev/pci/musycc_obsd.c |2 +-
1 files changed, 1 insertions(+), 1 deletions(-)
diff --git sys/dev/pci/musycc_obsd.c sys/dev/pci/musycc_obsd.c
index 25a58d8..0844136 100644
--- sys/dev/pci/musycc_obsd.c
+++ sys/dev/pci/musycc_obsd.c
@@ -215,7 +215,7 @@ mu
On Wed, 3 Jul 2013, Mark Kettenis wrote:
> > diff --git sys/arch/i386/i386/esm.c sys/arch/i386/i386/esm.c
> > index c90b2c4..3dff69e 100644
> > --- sys/arch/i386/i386/esm.c
> > +++ sys/arch/i386/i386/esm.c
> > @@ -880,7 +880,7 @@ esm_make_sensors(struct esm_softc *sc, struct
> > esm_devmap *devmap
On Wed, 3 Jul 2013, Mark Kettenis wrote:
> > diff --git sys/arch/i386/i386/db_interface.c
> > sys/arch/i386/i386/db_interface.c
> > index 85c1ff5..c75fd89 100644
> > --- sys/arch/i386/i386/db_interface.c
> > +++ sys/arch/i386/i386/db_interface.c
> > @@ -197,11 +197,11 @@ db_sysregs_cmd(db_expr_t a
Use %z* for size_t
while there, fix a few %d into %u
---
sys/arch/i386/pci/pcibios.c |2 +-
sys/dev/ic/bwi.c|4 ++--
sys/dev/ic/pgt.c|6 +++---
sys/dev/ic/ti.c |4 ++--
sys/dev/pci/amdiic.c|4 ++--
sys/dev/pci/amdpm.c |2
> Date: Wed, 3 Jul 2013 16:55:46 +0200 (CEST)
> From: Stefan Fritsch
>
> format string fixes: long
>
> ---
> sys/arch/i386/i386/esm.c |2 +-
> sys/kern/kern_descrip.c |4 ++--
> 2 files changed, 3 insertions(+), 3 deletions(-)
>
> diff --git sys/arch/i386/i386/esm.c sys/arch/i386/i386
> Date: Wed, 3 Jul 2013 16:51:43 +0200 (CEST)
> From: Stefan Fritsch
>
> fix: cast void * to char *, fix format args for (u)int64_t
>
> (tc_frequency is unsigned)
>
> diff --git sys/arch/i386/i386/db_disasm.c sys/arch/i386/i386/db_disasm.c
> index c9446a0..e5b9f4c 100644
> --- sys/a
On Wed, 3 Jul 2013, Mark Kettenis wrote:
> > Date: Wed, 3 Jul 2013 16:40:17 +0200 (CEST)
> > From: Stefan Fritsch
> >
> > don't pass empty format string in subr_disk.c
> >
> > this is necessary to enable -Wformat or -Wno-error=format
>
> Don't think this one makes much sense. Better to just do
cast time_t to long long (time_t is signed)
tv_nsec is long
---
sys/dev/vmt.c |4 ++--
sys/miscfs/procfs/procfs_status.c |4 ++--
2 files changed, 4 insertions(+), 4 deletions(-)
diff --git sys/dev/vmt.c sys/dev/vmt.c
index 5fb5e1e..34f1aec 100644
--- sys/dev/vmt.c
++
format string fixes: long
---
sys/arch/i386/i386/esm.c |2 +-
sys/kern/kern_descrip.c |4 ++--
2 files changed, 3 insertions(+), 3 deletions(-)
diff --git sys/arch/i386/i386/esm.c sys/arch/i386/i386/esm.c
index c90b2c4..3dff69e 100644
--- sys/arch/i386/i386/esm.c
+++ sys/arch/i386/i386/
fix: cast void * to char *, fix format args for (u)int64_t
(tc_frequency is unsigned)
diff --git sys/arch/i386/i386/db_disasm.c sys/arch/i386/i386/db_disasm.c
index c9446a0..e5b9f4c 100644
--- sys/arch/i386/i386/db_disasm.c
+++ sys/arch/i386/i386/db_disasm.c
@@ -1192,7 +1192,7 @@ db_d
> Date: Wed, 3 Jul 2013 16:40:17 +0200 (CEST)
> From: Stefan Fritsch
>
> don't pass empty format string in subr_disk.c
>
> this is necessary to enable -Wformat or -Wno-error=format
Don't think this one makes much sense. Better to just do:
log(pri, "%s", "");
and keep the rest of the code a
> Date: Wed, 3 Jul 2013 16:35:24 +0200 (CEST)
> From: Stefan Fritsch
>
> add support for %td for ptrdiff_t in kernel
>
> this also adds support in gcc 4.x kprintf
I'm on the fence about the CTASSERT here. If we ever support a code
model that's not ILP32 or LP64, we need a major overhaul of the
don't pass empty format string in subr_disk.c
this is necessary to enable -Wformat or -Wno-error=format
---
sys/kern/subr_disk.c | 11 ++-
1 files changed, 6 insertions(+), 5 deletions(-)
diff --git sys/kern/subr_disk.c sys/kern/subr_disk.c
index 4b500c1..2b1036a 100644
--- sys/kern/su
add support for %td for ptrdiff_t in kernel
this also adds support in gcc 4.x kprintf
---
gnu/gcc/gcc/c-format.c |7 ---
sys/kern/subr_prf.c|6 ++
2 files changed, 10 insertions(+), 3 deletions(-)
diff --git gnu/gcc/gcc/c-format.c gnu/gcc/gcc/c-format.c
index b9eecee..1b1734b
On Wednesday 03 July 2013, Alexander Bluhm wrote:
> On Wed, Jul 03, 2013 at 01:46:11PM +0200, Stefan Fritsch wrote:
> > which the patches are applied does not matter. Therefore I
> > thought that people could either review by patch from the above
> > directory or by source file from the combined pa
Please always include diffs in the emails; don't force us to go
to a browser.
On Wed, Jul 03, 2013 at 01:46:11PM +0200, Stefan Fritsch wrote:
> which the patches are applied does not matter. Therefore I thought that
> people could either review by patch from the above directory or by source
> file from the combined patch, just as they are comfortable. I would then
> commi
> Date: Wed, 3 Jul 2013 13:46:11 +0200 (CEST)
> From: Stefan Fritsch
>
> Hi,
>
> I have updated my patchset to enable format string warnings in kernel
> builds. Changes from the first version:
>
> - Regarding pointer differences, this now takes the approach favored by
> kettenis@ and adds sup
Hi,
I have updated my patchset to enable format string warnings in kernel
builds. Changes from the first version:
- Regarding pointer differences, this now takes the approach favored by
kettenis@ and adds support for %td in the kernel but does not add a
typedef for ptrdiff_t.
- It now removes
On Sun, 28 Apr 2013, Stefan Fritsch wrote:
> On Sun, 21 Apr 2013, Stefan Fritsch wrote:
> > On Sunday 21 April 2013, Philip Guenther wrote:
> > > Maybe this should be handled the bool bits in ,
> > > wrapping them in #ifdef _KERNEL ?
> >
> > #ifdef _KERNEL is fine with me.
>
> OK for the patch b
On Sunday 28 April 2013, Mark Kettenis wrote:
> Just drop the ptrdiff_t bit completely please. I
> don't think it should be used in the kernel; there always is a
> more appropriate kernel type to use.
Does it then still make sense to use %td in printf for differences of
pointers, or should we s
> Date: Sun, 28 Apr 2013 11:52:47 +0200 (CEST)
> From: Stefan Fritsch
>
> On Sun, 21 Apr 2013, Stefan Fritsch wrote:
> > On Sunday 21 April 2013, Philip Guenther wrote:
> >> Maybe this should be handled the bool bits in ,
> >> wrapping them in #ifdef _KERNEL ?
> >
> > #ifdef _KERNEL is fine with
On Sun, 21 Apr 2013, Stefan Fritsch wrote:
On Sunday 21 April 2013, Philip Guenther wrote:
Maybe this should be handled the bool bits in ,
wrapping them in #ifdef _KERNEL ?
#ifdef _KERNEL is fine with me.
OK for the patch below?
--- a/gnu/gcc/gcc/c-format.c
+++ b/gnu/gcc/gcc/c-format.c
@@ -
On Sunday 21 April 2013, Philip Guenther wrote:
> On Sun, Apr 21, 2013 at 12:15 PM, Stefan Fritsch
wrote:
> > On Sunday 21 April 2013, Mark Kettenis wrote:
> ...
>
> >> You can't put ptrdiff_t in either.
> >
> > Why not? POSIX only speaks about "at least the following types"
> > in sys/types.h
On Sun, Apr 21, 2013 at 12:15 PM, Stefan Fritsch wrote:
> On Sunday 21 April 2013, Mark Kettenis wrote:
...
>> You can't put ptrdiff_t in either.
>
> Why not? POSIX only speaks about "at least the following types" in
> sys/types.h, so there is no reason why it may not define more types.
Well, it
On Sunday 21 April 2013, Mark Kettenis wrote:
> > +++ b/sys/sys/types.h
> > @@ -177,6 +177,11 @@ typedef__clockid_t clockid_t;
> >
> > typedef __size_tsize_t;
> > #endif
> >
> > +#ifndef_PTRDIFF_T_DEFINED_
> > +#define_PTRDIFF_T_DEFINED_
> > +typedef__ptrdif
> Date: Sun, 21 Apr 2013 15:02:49 +0200 (CEST)
> From: Stefan Fritsch
>
> Moved ptrdiff_t definition to types.h with proper guards. Assume
> ptrdiff_t is long in subr_prf.c
>
> OK?
No.
> --- a/sys/sys/types.h
> +++ b/sys/sys/types.h
> @@ -177,6 +177,11 @@ typedef __clockid_t clockid_t;
>
Moved ptrdiff_t definition to types.h with proper guards. Assume
ptrdiff_t is long in subr_prf.c
OK?
--- a/gnu/gcc/gcc/c-format.c
+++ b/gnu/gcc/gcc/c-format.c
@@ -325,6 +325,7 @@ static const format_length_info kprintf_length_specs[] =
{ "l", FMT_LEN_l, STD_C89, "ll", FMT_LEN_ll, STD_C9L },
> Date: Sat, 20 Apr 2013 13:27:34 +0200 (CEST)
> From: Stefan Fritsch
>
> Adding ptrdiff_t to stdint.h is not strictly necessary, but I see no
> reason not to do it.
is and the C standard doesn't allow you to
have ptrdiff_t in
> I can't see that we would add an arch where sizeof(ptrdiff_t) !=
> sizeof(size_t).
We run on I32LP64 and ILP32, therefore, it seems better to match
't' to some other LONG print block, like 'D'.
size_t is an high-level abstraction, not a base type.
On Sun, 14 Apr 2013, Theo de Raadt wrote:
Since the kernel's printf does not support %td for ptrdiff_t, I have
used %ld instead. %zd would also work. Is there a preferred way?
Or try to add %td support to kernel printf?
The patch below seems to do the trick for gcc 4.2.
I can't see that we w
> Or LL128 ...
> But if such architectures come around, all kinds of things will need to be
> changed. This should not prevent anyone from documenting the current best
> practices.
All kinds of things will need to be changed? Quite the understatement.
On Sun, 14 Apr 2013, Theo de Raadt wrote:
That wasn't clear from Miod's response. So the policy is to assume
that char/short/int/long long are 8/16/32/64 bits and that intptr_t
is long?
We only run on C8S16I32L32P32 and C8S16I32L64P64 architectures. Short
names ILP32 and I32LP64. The additio
> > That wasn't clear from Miod's response. So the policy is to assume
> > that char/short/int/long long are 8/16/32/64 bits and that intptr_t
> > is long?
We only run on C8S16I32L32P32 and C8S16I32L64P64 architectures. Short
names ILP32 and I32LP64. The addition of any other sizing would
requ
On Saturday 06 April 2013, Stefan Fritsch wrote:
> That wasn't clear from Miod's response. So the policy is to assume
> that char/short/int/long long are 8/16/32/64 bits and that intptr_t
> is long? Should that be added to style(9), then?
I assume that the lack of a response means yes.
The patc
On Saturday 06 April 2013, Mark Kettenis wrote:
> > From: Stefan Fritsch
> > Date: Sat, 6 Apr 2013 13:04:04 +0200
> >
> > Also, there does not seem to be a header in the kernel yet, that
> > defines the standard PRI*. Should there be a new inttypes.h or,
> > since we probably want to define these
> From: Stefan Fritsch
> Date: Sat, 6 Apr 2013 13:04:04 +0200
>
> Also, there does not seem to be a header in the kernel yet, that
> defines the standard PRI*. Should there be a new inttypes.h or, since
> we probably want to define these always, do we put them into
> {,_}types.h?
There is no
On Saturday 06 April 2013, Miod Vallat wrote:
> > - define a few macros for special types, e.g. bus_addr_t and
> > time_t and use them
>
> It was decided to cast bus_*_t to long long and print them with
> %lld, rather than introduce kernel-only PRIfoo macros.
Is this documented somewhere? If not,
> compiling the kernel with -Wno-format is annoying for development and
> hides real bugs. It seems gcc has attribute format kprintf since a few
> years already. What is missing until we can remove -Wno-format? Any
> pointers to prior discussions about this?
It is missing the actual format fixe
Hi,
compiling the kernel with -Wno-format is annoying for development and
hides real bugs. It seems gcc has attribute format kprintf since a few
years already. What is missing until we can remove -Wno-format? Any
pointers to prior discussions about this?
AFAICS, things that remain to be done a
48 matches
Mail list logo
