Re: [git pull] vfs.git part 3

[Al Viro]

On Fri, Jul 07, 2017 at 04:00:02PM +0200, Christoph Hellwig wrote:
> On Fri, Jul 07, 2017 at 12:27:49AM +0100, Al Viro wrote:
> > On Thu, Jul 06, 2017 at 11:44:49PM +0200, Christoph Hellwig wrote:
> > > > Which sparse version are you using and what's your .config?
> > > 
> > > sparse is v0.5.0-62-gce18a90, .config is attached.
> > 
> > Arrgh...  OK, I see what's going on.  sparse commit affecting that
> > is "Allow casting to a restricted type if !restricted_value"; it
> > allows the things like (__le32)0.  It's present in sparse.git,
> > but not in chrisl/sparse.git, which is what you are using.
> 
> So what's the current story on sparse versions to use and releases?
> 
> At some point it seemed like upstream sparse was sort of dead
> and the chrisl repo was the one to use.  It seems the main sparse
> repo on git.kernel.org is now identical to the chrisl one, and he
> is doing the releases.  So maybe I'll just need to upgrade..

OK, here's what I have on top of #for-linus.  Handling of __bitwise in
{COMPAT_,}SYSCALL_DEFINE went into the tip of vfs.git#for-linus, works
both for old and for new sparse versions.  I'd added annotations of
sys_pwritev2(), etc., in syscalls.h and compat.h, other than that it's
what you'd posted...  Care to put your S-o-b on that?

commit 1c0ecce04fc5747dbd0578360895fe910d3124db
Author: Christoph Hellwig 
Date:   Thu Jul 6 18:58:37 2017 +0200

annotate RWF_... flags

[AV: added missing annotations in syscalls.h/compat.h]

Signed-off-by: Al Viro 

diff --git a/fs/nfsd/vfs.c b/fs/nfsd/vfs.c
index 38d0383dc7f9..bc69d40c4e8b 100644
--- a/fs/nfsd/vfs.c
+++ b/fs/nfsd/vfs.c
@@ -969,7 +969,7 @@ nfsd_vfs_write(struct svc_rqst *rqstp, struct svc_fh *fhp, 
struct file *file,
int use_wgather;
loff_t  pos = offset;
unsigned intpflags = current->flags;
-   int flags = 0;
+   rwf_t   flags = 0;
 
if (test_bit(RQ_LOCAL, >rq_flags))
/*
diff --git a/fs/read_write.c b/fs/read_write.c
index a2cbc8303dae..327d4aeefca0 100644
--- a/fs/read_write.c
+++ b/fs/read_write.c
@@ -33,7 +33,7 @@ const struct file_operations generic_ro_fops = {
 
 EXPORT_SYMBOL(generic_ro_fops);
 
-static inline int unsigned_offsets(struct file *file)
+static inline bool unsigned_offsets(struct file *file)
 {
return file->f_mode & FMODE_UNSIGNED_OFFSET;
 }
@@ -633,7 +633,7 @@ unsigned long iov_shorten(struct iovec *iov, unsigned long 
nr_segs, size_t to)
 EXPORT_SYMBOL(iov_shorten);
 
 static ssize_t do_iter_readv_writev(struct file *filp, struct iov_iter *iter,
-   loff_t *ppos, int type, int flags)
+   loff_t *ppos, int type, rwf_t flags)
 {
struct kiocb kiocb;
ssize_t ret;
@@ -655,7 +655,7 @@ static ssize_t do_iter_readv_writev(struct file *filp, 
struct iov_iter *iter,
 
 /* Do it by hand, with file-ops */
 static ssize_t do_loop_readv_writev(struct file *filp, struct iov_iter *iter,
-   loff_t *ppos, int type, int flags)
+   loff_t *ppos, int type, rwf_t flags)
 {
ssize_t ret = 0;
 
@@ -871,7 +871,7 @@ ssize_t compat_rw_copy_check_uvector(int type,
 #endif
 
 static ssize_t do_iter_read(struct file *file, struct iov_iter *iter,
-   loff_t *pos, int flags)
+   loff_t *pos, rwf_t flags)
 {
size_t tot_len;
ssize_t ret = 0;
@@ -899,7 +899,7 @@ static ssize_t do_iter_read(struct file *file, struct 
iov_iter *iter,
 }
 
 ssize_t vfs_iter_read(struct file *file, struct iov_iter *iter, loff_t *ppos,
-   int flags)
+   rwf_t flags)
 {
if (!file->f_op->read_iter)
return -EINVAL;
@@ -908,7 +908,7 @@ ssize_t vfs_iter_read(struct file *file, struct iov_iter 
*iter, loff_t *ppos,
 EXPORT_SYMBOL(vfs_iter_read);
 
 static ssize_t do_iter_write(struct file *file, struct iov_iter *iter,
-   loff_t *pos, int flags)
+   loff_t *pos, rwf_t flags)
 {
size_t tot_len;
ssize_t ret = 0;
@@ -937,7 +937,7 @@ static ssize_t do_iter_write(struct file *file, struct 
iov_iter *iter,
 }
 
 ssize_t vfs_iter_write(struct file *file, struct iov_iter *iter, loff_t *ppos,
-   int flags)
+   rwf_t flags)
 {
if (!file->f_op->write_iter)
return -EINVAL;
@@ -946,7 +946,7 @@ ssize_t vfs_iter_write(struct file *file, struct iov_iter 
*iter, loff_t *ppos,
 EXPORT_SYMBOL(vfs_iter_write);
 
 ssize_t vfs_readv(struct file *file, const struct iovec __user *vec,
- unsigned long vlen, loff_t *pos, int flags)
+ unsigned long vlen, loff_t *pos, rwf_t flags)
 {
struct iovec iovstack[UIO_FASTIOV];
struct iovec *iov = iovstack;
@@ -964,7 +964,7 @@ ssize_t vfs_readv(struct file *file, const struct iovec 
__user *vec,
 EXPORT_SYMBOL(vfs_readv);
 
 ssize_t vfs_writev(struct file *file, 

Re: [git pull] vfs.git part 3

[Al Viro]

On Fri, Jul 07, 2017 at 04:00:02PM +0200, Christoph Hellwig wrote:
> On Fri, Jul 07, 2017 at 12:27:49AM +0100, Al Viro wrote:
> > On Thu, Jul 06, 2017 at 11:44:49PM +0200, Christoph Hellwig wrote:
> > > > Which sparse version are you using and what's your .config?
> > > 
> > > sparse is v0.5.0-62-gce18a90, .config is attached.
> > 
> > Arrgh...  OK, I see what's going on.  sparse commit affecting that
> > is "Allow casting to a restricted type if !restricted_value"; it
> > allows the things like (__le32)0.  It's present in sparse.git,
> > but not in chrisl/sparse.git, which is what you are using.
> 
> So what's the current story on sparse versions to use and releases?
> 
> At some point it seemed like upstream sparse was sort of dead
> and the chrisl repo was the one to use.  It seems the main sparse
> repo on git.kernel.org is now identical to the chrisl one, and he
> is doing the releases.  So maybe I'll just need to upgrade..

OK, here's what I have on top of #for-linus.  Handling of __bitwise in
{COMPAT_,}SYSCALL_DEFINE went into the tip of vfs.git#for-linus, works
both for old and for new sparse versions.  I'd added annotations of
sys_pwritev2(), etc., in syscalls.h and compat.h, other than that it's
what you'd posted...  Care to put your S-o-b on that?

commit 1c0ecce04fc5747dbd0578360895fe910d3124db
Author: Christoph Hellwig 
Date:   Thu Jul 6 18:58:37 2017 +0200

annotate RWF_... flags

[AV: added missing annotations in syscalls.h/compat.h]

Signed-off-by: Al Viro 

diff --git a/fs/nfsd/vfs.c b/fs/nfsd/vfs.c
index 38d0383dc7f9..bc69d40c4e8b 100644
--- a/fs/nfsd/vfs.c
+++ b/fs/nfsd/vfs.c
@@ -969,7 +969,7 @@ nfsd_vfs_write(struct svc_rqst *rqstp, struct svc_fh *fhp, 
struct file *file,
int use_wgather;
loff_t  pos = offset;
unsigned intpflags = current->flags;
-   int flags = 0;
+   rwf_t   flags = 0;
 
if (test_bit(RQ_LOCAL, >rq_flags))
/*
diff --git a/fs/read_write.c b/fs/read_write.c
index a2cbc8303dae..327d4aeefca0 100644
--- a/fs/read_write.c
+++ b/fs/read_write.c
@@ -33,7 +33,7 @@ const struct file_operations generic_ro_fops = {
 
 EXPORT_SYMBOL(generic_ro_fops);
 
-static inline int unsigned_offsets(struct file *file)
+static inline bool unsigned_offsets(struct file *file)
 {
return file->f_mode & FMODE_UNSIGNED_OFFSET;
 }
@@ -633,7 +633,7 @@ unsigned long iov_shorten(struct iovec *iov, unsigned long 
nr_segs, size_t to)
 EXPORT_SYMBOL(iov_shorten);
 
 static ssize_t do_iter_readv_writev(struct file *filp, struct iov_iter *iter,
-   loff_t *ppos, int type, int flags)
+   loff_t *ppos, int type, rwf_t flags)
 {
struct kiocb kiocb;
ssize_t ret;
@@ -655,7 +655,7 @@ static ssize_t do_iter_readv_writev(struct file *filp, 
struct iov_iter *iter,
 
 /* Do it by hand, with file-ops */
 static ssize_t do_loop_readv_writev(struct file *filp, struct iov_iter *iter,
-   loff_t *ppos, int type, int flags)
+   loff_t *ppos, int type, rwf_t flags)
 {
ssize_t ret = 0;
 
@@ -871,7 +871,7 @@ ssize_t compat_rw_copy_check_uvector(int type,
 #endif
 
 static ssize_t do_iter_read(struct file *file, struct iov_iter *iter,
-   loff_t *pos, int flags)
+   loff_t *pos, rwf_t flags)
 {
size_t tot_len;
ssize_t ret = 0;
@@ -899,7 +899,7 @@ static ssize_t do_iter_read(struct file *file, struct 
iov_iter *iter,
 }
 
 ssize_t vfs_iter_read(struct file *file, struct iov_iter *iter, loff_t *ppos,
-   int flags)
+   rwf_t flags)
 {
if (!file->f_op->read_iter)
return -EINVAL;
@@ -908,7 +908,7 @@ ssize_t vfs_iter_read(struct file *file, struct iov_iter 
*iter, loff_t *ppos,
 EXPORT_SYMBOL(vfs_iter_read);
 
 static ssize_t do_iter_write(struct file *file, struct iov_iter *iter,
-   loff_t *pos, int flags)
+   loff_t *pos, rwf_t flags)
 {
size_t tot_len;
ssize_t ret = 0;
@@ -937,7 +937,7 @@ static ssize_t do_iter_write(struct file *file, struct 
iov_iter *iter,
 }
 
 ssize_t vfs_iter_write(struct file *file, struct iov_iter *iter, loff_t *ppos,
-   int flags)
+   rwf_t flags)
 {
if (!file->f_op->write_iter)
return -EINVAL;
@@ -946,7 +946,7 @@ ssize_t vfs_iter_write(struct file *file, struct iov_iter 
*iter, loff_t *ppos,
 EXPORT_SYMBOL(vfs_iter_write);
 
 ssize_t vfs_readv(struct file *file, const struct iovec __user *vec,
- unsigned long vlen, loff_t *pos, int flags)
+ unsigned long vlen, loff_t *pos, rwf_t flags)
 {
struct iovec iovstack[UIO_FASTIOV];
struct iovec *iov = iovstack;
@@ -964,7 +964,7 @@ ssize_t vfs_readv(struct file *file, const struct iovec 
__user *vec,
 EXPORT_SYMBOL(vfs_readv);
 
 ssize_t vfs_writev(struct file *file, const struct iovec __user *vec,
-  

Re: [git pull] vfs.git part 3

[Christopher Li]

On Fri, Jul 7, 2017 at 8:48 AM, Linus Torvalds
 wrote:
> The releases are done way too seldom to be useful, but that may be
> improving. There is one fairly imminent, and it's probably a good idea
> to just test the current git tree.

Yes guilty of too few releases. We are cutting one release pretty soon.
The currently master branch of sparse git repository has gone to  RC4
of  of release v0.5.1.

BTW, the current sparse development has move back to the official sparse
repository. The sparse-next is the branch to test the bleeding edge bits.
In theory sparse-next can roll back and rewrite history, there is no guarantee
it will be clean pull. The master branch will never rebase, it will
prove a clean pull.

Chris

Re: [git pull] vfs.git part 3

[Christopher Li]

On Fri, Jul 7, 2017 at 8:48 AM, Linus Torvalds
 wrote:
> The releases are done way too seldom to be useful, but that may be
> improving. There is one fairly imminent, and it's probably a good idea
> to just test the current git tree.

Yes guilty of too few releases. We are cutting one release pretty soon.
The currently master branch of sparse git repository has gone to  RC4
of  of release v0.5.1.

BTW, the current sparse development has move back to the official sparse
repository. The sparse-next is the branch to test the bleeding edge bits.
In theory sparse-next can roll back and rewrite history, there is no guarantee
it will be clean pull. The master branch will never rebase, it will
prove a clean pull.

Chris

Re: [git pull] vfs.git part 3

[Linus Torvalds]

On Fri, Jul 7, 2017 at 7:00 AM, Christoph Hellwig  wrote:
>
> So what's the current story on sparse versions to use and releases?

The releases are done way too seldom to be useful, but that may be
improving. There is one fairly imminent, and it's probably a good idea
to just test the current git tree.

   Linus

Re: [git pull] vfs.git part 3

[Linus Torvalds]

On Fri, Jul 7, 2017 at 7:00 AM, Christoph Hellwig  wrote:
>
> So what's the current story on sparse versions to use and releases?

The releases are done way too seldom to be useful, but that may be
improving. There is one fairly imminent, and it's probably a good idea
to just test the current git tree.

   Linus

Re: [git pull] vfs.git part 3

[Christoph Hellwig]

On Fri, Jul 07, 2017 at 12:27:49AM +0100, Al Viro wrote:
> On Thu, Jul 06, 2017 at 11:44:49PM +0200, Christoph Hellwig wrote:
> > > Which sparse version are you using and what's your .config?
> > 
> > sparse is v0.5.0-62-gce18a90, .config is attached.
> 
> Arrgh...  OK, I see what's going on.  sparse commit affecting that
> is "Allow casting to a restricted type if !restricted_value"; it
> allows the things like (__le32)0.  It's present in sparse.git,
> but not in chrisl/sparse.git, which is what you are using.

So what's the current story on sparse versions to use and releases?

At some point it seemed like upstream sparse was sort of dead
and the chrisl repo was the one to use.  It seems the main sparse
repo on git.kernel.org is now identical to the chrisl one, and he
is doing the releases.  So maybe I'll just need to upgrade..

Re: [git pull] vfs.git part 3

[Christoph Hellwig]

On Fri, Jul 07, 2017 at 12:27:49AM +0100, Al Viro wrote:
> On Thu, Jul 06, 2017 at 11:44:49PM +0200, Christoph Hellwig wrote:
> > > Which sparse version are you using and what's your .config?
> > 
> > sparse is v0.5.0-62-gce18a90, .config is attached.
> 
> Arrgh...  OK, I see what's going on.  sparse commit affecting that
> is "Allow casting to a restricted type if !restricted_value"; it
> allows the things like (__le32)0.  It's present in sparse.git,
> but not in chrisl/sparse.git, which is what you are using.

So what's the current story on sparse versions to use and releases?

At some point it seemed like upstream sparse was sort of dead
and the chrisl repo was the one to use.  It seems the main sparse
repo on git.kernel.org is now identical to the chrisl one, and he
is doing the releases.  So maybe I'll just need to upgrade..

Re: [git pull] vfs.git part 3

[Al Viro]

On Thu, Jul 06, 2017 at 11:44:49PM +0200, Christoph Hellwig wrote:
> > Which sparse version are you using and what's your .config?
> 
> sparse is v0.5.0-62-gce18a90, .config is attached.

Arrgh...  OK, I see what's going on.  sparse commit affecting that
is "Allow casting to a restricted type if !restricted_value"; it
allows the things like (__le32)0.  It's present in sparse.git,
but not in chrisl/sparse.git, which is what you are using.

Anyway, the thing I'd missed kernel-side is this:
#define __TYPE_IS_L(t)  (__same_type((t)0, 0L))
#define __TYPE_IS_UL(t) (__same_type((t)0, 0UL))
#define __TYPE_IS_LL(t) (__same_type((t)0, 0LL) || __same_type((t)0, 0ULL))

Let's turn them into
#define __TYPE_AS(t, v) __same_type((__force t)0, v) 
#define __TYPE_IS_L(t)  (__TYPE_AS(t, 0L))
#define __TYPE_IS_UL(t) (__TYPE_AS(t, 0UL))
#define __TYPE_IS_LL(t) (__TYPE_AS(t, 0LL) || __TYPE_AS(t, 0ULL))

That should do it both for old and for new versions of sparse.

Re: [git pull] vfs.git part 3

[Al Viro]

On Thu, Jul 06, 2017 at 11:44:49PM +0200, Christoph Hellwig wrote:
> > Which sparse version are you using and what's your .config?
> 
> sparse is v0.5.0-62-gce18a90, .config is attached.

Arrgh...  OK, I see what's going on.  sparse commit affecting that
is "Allow casting to a restricted type if !restricted_value"; it
allows the things like (__le32)0.  It's present in sparse.git,
but not in chrisl/sparse.git, which is what you are using.

Anyway, the thing I'd missed kernel-side is this:
#define __TYPE_IS_L(t)  (__same_type((t)0, 0L))
#define __TYPE_IS_UL(t) (__same_type((t)0, 0UL))
#define __TYPE_IS_LL(t) (__same_type((t)0, 0LL) || __same_type((t)0, 0ULL))

Let's turn them into
#define __TYPE_AS(t, v) __same_type((__force t)0, v) 
#define __TYPE_IS_L(t)  (__TYPE_AS(t, 0L))
#define __TYPE_IS_UL(t) (__TYPE_AS(t, 0UL))
#define __TYPE_IS_LL(t) (__TYPE_AS(t, 0LL) || __TYPE_AS(t, 0ULL))

That should do it both for old and for new versions of sparse.

Re: [git pull] vfs.git part 3

[Christoph Hellwig]

> Which sparse version are you using and what's your .config?

sparse is v0.5.0-62-gce18a90, .config is attached.



.config
Description: application/config

Re: [git pull] vfs.git part 3

[Christoph Hellwig]

> Which sparse version are you using and what's your .config?

sparse is v0.5.0-62-gce18a90, .config is attached.



.config
Description: application/config

Re: [git pull] vfs.git part 3

[Al Viro]

On Thu, Jul 06, 2017 at 06:58:37PM +0200, Christoph Hellwig wrote:
> On Thu, Jul 06, 2017 at 04:51:13PM +0100, Al Viro wrote:
> > On Thu, Jul 06, 2017 at 04:46:02PM +0100, Al Viro wrote:
> > 
> > > That - on #work.read_write, as in vfs.git at the moment...
> > 
> > ... and for COMPAT_SYSCALL you need
> > #define __SC_DELOUSE(t,v) ((__force t)(unsigned long)(v))
> > in linux/compat.h
> 
> I'm still getting warnings with both these force casts.  This is
> the current stack:

That + Linus' tree as of the end of yesterday =>
  CHECK   fs/read_write.c
fs/read_write.c:38:29: warning: incorrect type in return expression (different 
base types)
fs/read_write.c:38:29:expected int
fs/read_write.c:38:29:got restricted fmode_t
fs/read_write.c:38:29: warning: incorrect type in return expression (different 
base types)
fs/read_write.c:38:29:expected int
fs/read_write.c:38:29:got restricted fmode_t
fs/read_write.c:38:29: warning: incorrect type in return expression (different 
base types)
fs/read_write.c:38:29:expected int
fs/read_write.c:38:29:got restricted fmode_t
fs/read_write.c:38:29: warning: incorrect type in return expression (different 
base types)
fs/read_write.c:38:29:expected int
fs/read_write.c:38:29:got restricted fmode_t

All of which are from unsigned_offsets() and that's one case where bool would 
be better than
int.  Switching the return type to bool yields
  CHECK   fs/read_write.c
  CC  fs/read_write.o
- no warnings at all.

Which sparse version are you using and what's your .config?

Re: [git pull] vfs.git part 3

[Al Viro]

On Thu, Jul 06, 2017 at 06:58:37PM +0200, Christoph Hellwig wrote:
> On Thu, Jul 06, 2017 at 04:51:13PM +0100, Al Viro wrote:
> > On Thu, Jul 06, 2017 at 04:46:02PM +0100, Al Viro wrote:
> > 
> > > That - on #work.read_write, as in vfs.git at the moment...
> > 
> > ... and for COMPAT_SYSCALL you need
> > #define __SC_DELOUSE(t,v) ((__force t)(unsigned long)(v))
> > in linux/compat.h
> 
> I'm still getting warnings with both these force casts.  This is
> the current stack:

That + Linus' tree as of the end of yesterday =>
  CHECK   fs/read_write.c
fs/read_write.c:38:29: warning: incorrect type in return expression (different 
base types)
fs/read_write.c:38:29:expected int
fs/read_write.c:38:29:got restricted fmode_t
fs/read_write.c:38:29: warning: incorrect type in return expression (different 
base types)
fs/read_write.c:38:29:expected int
fs/read_write.c:38:29:got restricted fmode_t
fs/read_write.c:38:29: warning: incorrect type in return expression (different 
base types)
fs/read_write.c:38:29:expected int
fs/read_write.c:38:29:got restricted fmode_t
fs/read_write.c:38:29: warning: incorrect type in return expression (different 
base types)
fs/read_write.c:38:29:expected int
fs/read_write.c:38:29:got restricted fmode_t

All of which are from unsigned_offsets() and that's one case where bool would 
be better than
int.  Switching the return type to bool yields
  CHECK   fs/read_write.c
  CC  fs/read_write.o
- no warnings at all.

Which sparse version are you using and what's your .config?

Re: [git pull] vfs.git part 3

[Christoph Hellwig]

On Thu, Jul 06, 2017 at 04:51:13PM +0100, Al Viro wrote:
> On Thu, Jul 06, 2017 at 04:46:02PM +0100, Al Viro wrote:
> 
> > That - on #work.read_write, as in vfs.git at the moment...
> 
> ... and for COMPAT_SYSCALL you need
> #define __SC_DELOUSE(t,v) ((__force t)(unsigned long)(v))
> in linux/compat.h

I'm still getting warnings with both these force casts.  This is
the current stack:


diff --git a/arch/s390/include/asm/compat.h b/arch/s390/include/asm/compat.h
index 0ddd37e6c29d..019f0de892c3 100644
--- a/arch/s390/include/asm/compat.h
+++ b/arch/s390/include/asm/compat.h
@@ -12,7 +12,7 @@
 
 #define __SC_DELOUSE(t,v) ({ \
BUILD_BUG_ON(sizeof(t) > 4 && !__TYPE_IS_PTR(t)); \
-   (t)(__TYPE_IS_PTR(t) ? ((v) & 0x7fff) : (v)); \
+   (__force t)(__TYPE_IS_PTR(t) ? ((v) & 0x7fff) : (v)); \
 })
 
 #define PSW32_MASK_PER 0x4000UL
diff --git a/fs/nfsd/vfs.c b/fs/nfsd/vfs.c
index 38d0383dc7f9..bc69d40c4e8b 100644
--- a/fs/nfsd/vfs.c
+++ b/fs/nfsd/vfs.c
@@ -969,7 +969,7 @@ nfsd_vfs_write(struct svc_rqst *rqstp, struct svc_fh *fhp, 
struct file *file,
int use_wgather;
loff_t  pos = offset;
unsigned intpflags = current->flags;
-   int flags = 0;
+   rwf_t   flags = 0;
 
if (test_bit(RQ_LOCAL, >rq_flags))
/*
diff --git a/fs/read_write.c b/fs/read_write.c
index a2cbc8303dae..bc2db5e5cd19 100644
--- a/fs/read_write.c
+++ b/fs/read_write.c
@@ -633,7 +633,7 @@ unsigned long iov_shorten(struct iovec *iov, unsigned long 
nr_segs, size_t to)
 EXPORT_SYMBOL(iov_shorten);
 
 static ssize_t do_iter_readv_writev(struct file *filp, struct iov_iter *iter,
-   loff_t *ppos, int type, int flags)
+   loff_t *ppos, int type, rwf_t flags)
 {
struct kiocb kiocb;
ssize_t ret;
@@ -655,7 +655,7 @@ static ssize_t do_iter_readv_writev(struct file *filp, 
struct iov_iter *iter,
 
 /* Do it by hand, with file-ops */
 static ssize_t do_loop_readv_writev(struct file *filp, struct iov_iter *iter,
-   loff_t *ppos, int type, int flags)
+   loff_t *ppos, int type, rwf_t flags)
 {
ssize_t ret = 0;
 
@@ -871,7 +871,7 @@ ssize_t compat_rw_copy_check_uvector(int type,
 #endif
 
 static ssize_t do_iter_read(struct file *file, struct iov_iter *iter,
-   loff_t *pos, int flags)
+   loff_t *pos, rwf_t flags)
 {
size_t tot_len;
ssize_t ret = 0;
@@ -899,7 +899,7 @@ static ssize_t do_iter_read(struct file *file, struct 
iov_iter *iter,
 }
 
 ssize_t vfs_iter_read(struct file *file, struct iov_iter *iter, loff_t *ppos,
-   int flags)
+   rwf_t flags)
 {
if (!file->f_op->read_iter)
return -EINVAL;
@@ -908,7 +908,7 @@ ssize_t vfs_iter_read(struct file *file, struct iov_iter 
*iter, loff_t *ppos,
 EXPORT_SYMBOL(vfs_iter_read);
 
 static ssize_t do_iter_write(struct file *file, struct iov_iter *iter,
-   loff_t *pos, int flags)
+   loff_t *pos, rwf_t flags)
 {
size_t tot_len;
ssize_t ret = 0;
@@ -937,7 +937,7 @@ static ssize_t do_iter_write(struct file *file, struct 
iov_iter *iter,
 }
 
 ssize_t vfs_iter_write(struct file *file, struct iov_iter *iter, loff_t *ppos,
-   int flags)
+   rwf_t flags)
 {
if (!file->f_op->write_iter)
return -EINVAL;
@@ -946,7 +946,7 @@ ssize_t vfs_iter_write(struct file *file, struct iov_iter 
*iter, loff_t *ppos,
 EXPORT_SYMBOL(vfs_iter_write);
 
 ssize_t vfs_readv(struct file *file, const struct iovec __user *vec,
- unsigned long vlen, loff_t *pos, int flags)
+ unsigned long vlen, loff_t *pos, rwf_t flags)
 {
struct iovec iovstack[UIO_FASTIOV];
struct iovec *iov = iovstack;
@@ -964,7 +964,7 @@ ssize_t vfs_readv(struct file *file, const struct iovec 
__user *vec,
 EXPORT_SYMBOL(vfs_readv);
 
 ssize_t vfs_writev(struct file *file, const struct iovec __user *vec,
-  unsigned long vlen, loff_t *pos, int flags)
+  unsigned long vlen, loff_t *pos, rwf_t flags)
 {
struct iovec iovstack[UIO_FASTIOV];
struct iovec *iov = iovstack;
@@ -981,7 +981,7 @@ ssize_t vfs_writev(struct file *file, const struct iovec 
__user *vec,
 EXPORT_SYMBOL(vfs_writev);
 
 static ssize_t do_readv(unsigned long fd, const struct iovec __user *vec,
-   unsigned long vlen, int flags)
+   unsigned long vlen, rwf_t flags)
 {
struct fd f = fdget_pos(fd);
ssize_t ret = -EBADF;
@@ -1001,7 +1001,7 @@ static ssize_t do_readv(unsigned long fd, const struct 
iovec __user *vec,
 }
 
 static ssize_t do_writev(unsigned long fd, const struct iovec __user *vec,
-unsigned long vlen, int flags)
+unsigned long vlen, rwf_t flags)
 {
struct fd f = fdget_pos(fd);
 

Re: [git pull] vfs.git part 3

[Christoph Hellwig]

On Thu, Jul 06, 2017 at 04:51:13PM +0100, Al Viro wrote:
> On Thu, Jul 06, 2017 at 04:46:02PM +0100, Al Viro wrote:
> 
> > That - on #work.read_write, as in vfs.git at the moment...
> 
> ... and for COMPAT_SYSCALL you need
> #define __SC_DELOUSE(t,v) ((__force t)(unsigned long)(v))
> in linux/compat.h

I'm still getting warnings with both these force casts.  This is
the current stack:


diff --git a/arch/s390/include/asm/compat.h b/arch/s390/include/asm/compat.h
index 0ddd37e6c29d..019f0de892c3 100644
--- a/arch/s390/include/asm/compat.h
+++ b/arch/s390/include/asm/compat.h
@@ -12,7 +12,7 @@
 
 #define __SC_DELOUSE(t,v) ({ \
BUILD_BUG_ON(sizeof(t) > 4 && !__TYPE_IS_PTR(t)); \
-   (t)(__TYPE_IS_PTR(t) ? ((v) & 0x7fff) : (v)); \
+   (__force t)(__TYPE_IS_PTR(t) ? ((v) & 0x7fff) : (v)); \
 })
 
 #define PSW32_MASK_PER 0x4000UL
diff --git a/fs/nfsd/vfs.c b/fs/nfsd/vfs.c
index 38d0383dc7f9..bc69d40c4e8b 100644
--- a/fs/nfsd/vfs.c
+++ b/fs/nfsd/vfs.c
@@ -969,7 +969,7 @@ nfsd_vfs_write(struct svc_rqst *rqstp, struct svc_fh *fhp, 
struct file *file,
int use_wgather;
loff_t  pos = offset;
unsigned intpflags = current->flags;
-   int flags = 0;
+   rwf_t   flags = 0;
 
if (test_bit(RQ_LOCAL, >rq_flags))
/*
diff --git a/fs/read_write.c b/fs/read_write.c
index a2cbc8303dae..bc2db5e5cd19 100644
--- a/fs/read_write.c
+++ b/fs/read_write.c
@@ -633,7 +633,7 @@ unsigned long iov_shorten(struct iovec *iov, unsigned long 
nr_segs, size_t to)
 EXPORT_SYMBOL(iov_shorten);
 
 static ssize_t do_iter_readv_writev(struct file *filp, struct iov_iter *iter,
-   loff_t *ppos, int type, int flags)
+   loff_t *ppos, int type, rwf_t flags)
 {
struct kiocb kiocb;
ssize_t ret;
@@ -655,7 +655,7 @@ static ssize_t do_iter_readv_writev(struct file *filp, 
struct iov_iter *iter,
 
 /* Do it by hand, with file-ops */
 static ssize_t do_loop_readv_writev(struct file *filp, struct iov_iter *iter,
-   loff_t *ppos, int type, int flags)
+   loff_t *ppos, int type, rwf_t flags)
 {
ssize_t ret = 0;
 
@@ -871,7 +871,7 @@ ssize_t compat_rw_copy_check_uvector(int type,
 #endif
 
 static ssize_t do_iter_read(struct file *file, struct iov_iter *iter,
-   loff_t *pos, int flags)
+   loff_t *pos, rwf_t flags)
 {
size_t tot_len;
ssize_t ret = 0;
@@ -899,7 +899,7 @@ static ssize_t do_iter_read(struct file *file, struct 
iov_iter *iter,
 }
 
 ssize_t vfs_iter_read(struct file *file, struct iov_iter *iter, loff_t *ppos,
-   int flags)
+   rwf_t flags)
 {
if (!file->f_op->read_iter)
return -EINVAL;
@@ -908,7 +908,7 @@ ssize_t vfs_iter_read(struct file *file, struct iov_iter 
*iter, loff_t *ppos,
 EXPORT_SYMBOL(vfs_iter_read);
 
 static ssize_t do_iter_write(struct file *file, struct iov_iter *iter,
-   loff_t *pos, int flags)
+   loff_t *pos, rwf_t flags)
 {
size_t tot_len;
ssize_t ret = 0;
@@ -937,7 +937,7 @@ static ssize_t do_iter_write(struct file *file, struct 
iov_iter *iter,
 }
 
 ssize_t vfs_iter_write(struct file *file, struct iov_iter *iter, loff_t *ppos,
-   int flags)
+   rwf_t flags)
 {
if (!file->f_op->write_iter)
return -EINVAL;
@@ -946,7 +946,7 @@ ssize_t vfs_iter_write(struct file *file, struct iov_iter 
*iter, loff_t *ppos,
 EXPORT_SYMBOL(vfs_iter_write);
 
 ssize_t vfs_readv(struct file *file, const struct iovec __user *vec,
- unsigned long vlen, loff_t *pos, int flags)
+ unsigned long vlen, loff_t *pos, rwf_t flags)
 {
struct iovec iovstack[UIO_FASTIOV];
struct iovec *iov = iovstack;
@@ -964,7 +964,7 @@ ssize_t vfs_readv(struct file *file, const struct iovec 
__user *vec,
 EXPORT_SYMBOL(vfs_readv);
 
 ssize_t vfs_writev(struct file *file, const struct iovec __user *vec,
-  unsigned long vlen, loff_t *pos, int flags)
+  unsigned long vlen, loff_t *pos, rwf_t flags)
 {
struct iovec iovstack[UIO_FASTIOV];
struct iovec *iov = iovstack;
@@ -981,7 +981,7 @@ ssize_t vfs_writev(struct file *file, const struct iovec 
__user *vec,
 EXPORT_SYMBOL(vfs_writev);
 
 static ssize_t do_readv(unsigned long fd, const struct iovec __user *vec,
-   unsigned long vlen, int flags)
+   unsigned long vlen, rwf_t flags)
 {
struct fd f = fdget_pos(fd);
ssize_t ret = -EBADF;
@@ -1001,7 +1001,7 @@ static ssize_t do_readv(unsigned long fd, const struct 
iovec __user *vec,
 }
 
 static ssize_t do_writev(unsigned long fd, const struct iovec __user *vec,
-unsigned long vlen, int flags)
+unsigned long vlen, rwf_t flags)
 {
struct fd f = fdget_pos(fd);
 

Re: [git pull] vfs.git part 3

[Al Viro]

On Thu, Jul 06, 2017 at 04:46:02PM +0100, Al Viro wrote:

> That - on #work.read_write, as in vfs.git at the moment...

... and for COMPAT_SYSCALL you need
#define __SC_DELOUSE(t,v) ((__force t)(unsigned long)(v))
in linux/compat.h

Re: [git pull] vfs.git part 3

[Al Viro]

On Thu, Jul 06, 2017 at 04:46:02PM +0100, Al Viro wrote:

> That - on #work.read_write, as in vfs.git at the moment...

... and for COMPAT_SYSCALL you need
#define __SC_DELOUSE(t,v) ((__force t)(unsigned long)(v))
in linux/compat.h

Re: [git pull] vfs.git part 3

[Al Viro]

On Thu, Jul 06, 2017 at 05:10:33PM +0200, Christoph Hellwig wrote:
> On Thu, Jul 06, 2017 at 04:03:30PM +0100, Al Viro wrote:
> > On Thu, Jul 06, 2017 at 04:48:40PM +0200, Christoph Hellwig wrote:
> > 
> > > Just did the whole batch (patch below), but it seems like using a
> > > __bitwise type in SYSCALL_DEFINE* will always give warnings like:
> > > 
> > > fs/read_write.c:1095:1: warning: cast to restricted __kernel_rwf_t
> > > 
> > > which I'm not sure to deal with..
> > 
> > #define __SC_CAST(t, a) (__force t) a
> 
> doesn't seem to make a difference..

Works here...

; cat a.c
#include 
#undef __SC_CAST
#define __SC_CAST(t, a) (__force t)a
typedef int __bitwise __foo_t;
static __foo_t is_OK;
static int will_warn;
SYSCALL_DEFINE1(foo, __foo_t, arg)
{
is_OK = arg;
will_warn = arg;
return 0;
}
; make a.o C=2 CHECK=~/local/sparse/sparse 
  CHK include/config/kernel.release
  CHK include/generated/uapi/linux/version.h
  CHK include/generated/utsrelease.h
  CHECK   arch/x86/purgatory/purgatory.c
  CHECK   arch/x86/purgatory/sha256.c
  CHECK   arch/x86/purgatory/string.c
arch/x86/purgatory/../boot/string.c:165:6: warning: symbol 'strchr' was not 
declared. Should it be static?
  CHK include/generated/bounds.h
  CHK include/generated/timeconst.h
  CHK include/generated/asm-offsets.h
  CALLscripts/checksyscalls.sh
  DESCEND  objtool
  CHECK   scripts/mod/empty.c
  CHK scripts/mod/devicetable-offsets.h
  CHK include/generated/timeconst.h
  CHK include/generated/bounds.h
  CHECK   a.c
a.c:10:19: warning: incorrect type in assignment (different base types)
a.c:10:19:expected int static [signed] [toplevel] will_warn
a.c:10:19:got restricted __foo_t [usertype] arg
  CC  a.o
;

That - on #work.read_write, as in vfs.git at the moment...

Re: [git pull] vfs.git part 3

[Al Viro]

On Thu, Jul 06, 2017 at 05:10:33PM +0200, Christoph Hellwig wrote:
> On Thu, Jul 06, 2017 at 04:03:30PM +0100, Al Viro wrote:
> > On Thu, Jul 06, 2017 at 04:48:40PM +0200, Christoph Hellwig wrote:
> > 
> > > Just did the whole batch (patch below), but it seems like using a
> > > __bitwise type in SYSCALL_DEFINE* will always give warnings like:
> > > 
> > > fs/read_write.c:1095:1: warning: cast to restricted __kernel_rwf_t
> > > 
> > > which I'm not sure to deal with..
> > 
> > #define __SC_CAST(t, a) (__force t) a
> 
> doesn't seem to make a difference..

Works here...

; cat a.c
#include 
#undef __SC_CAST
#define __SC_CAST(t, a) (__force t)a
typedef int __bitwise __foo_t;
static __foo_t is_OK;
static int will_warn;
SYSCALL_DEFINE1(foo, __foo_t, arg)
{
is_OK = arg;
will_warn = arg;
return 0;
}
; make a.o C=2 CHECK=~/local/sparse/sparse 
  CHK include/config/kernel.release
  CHK include/generated/uapi/linux/version.h
  CHK include/generated/utsrelease.h
  CHECK   arch/x86/purgatory/purgatory.c
  CHECK   arch/x86/purgatory/sha256.c
  CHECK   arch/x86/purgatory/string.c
arch/x86/purgatory/../boot/string.c:165:6: warning: symbol 'strchr' was not 
declared. Should it be static?
  CHK include/generated/bounds.h
  CHK include/generated/timeconst.h
  CHK include/generated/asm-offsets.h
  CALLscripts/checksyscalls.sh
  DESCEND  objtool
  CHECK   scripts/mod/empty.c
  CHK scripts/mod/devicetable-offsets.h
  CHK include/generated/timeconst.h
  CHK include/generated/bounds.h
  CHECK   a.c
a.c:10:19: warning: incorrect type in assignment (different base types)
a.c:10:19:expected int static [signed] [toplevel] will_warn
a.c:10:19:got restricted __foo_t [usertype] arg
  CC  a.o
;

That - on #work.read_write, as in vfs.git at the moment...

Re: [git pull] vfs.git part 3

[Christoph Hellwig]

On Thu, Jul 06, 2017 at 04:03:30PM +0100, Al Viro wrote:
> On Thu, Jul 06, 2017 at 04:48:40PM +0200, Christoph Hellwig wrote:
> 
> > Just did the whole batch (patch below), but it seems like using a
> > __bitwise type in SYSCALL_DEFINE* will always give warnings like:
> > 
> > fs/read_write.c:1095:1: warning: cast to restricted __kernel_rwf_t
> > 
> > which I'm not sure to deal with..
> 
> #define __SC_CAST(t, a) (__force t) a

doesn't seem to make a difference..

Re: [git pull] vfs.git part 3

[Christoph Hellwig]

On Thu, Jul 06, 2017 at 04:03:30PM +0100, Al Viro wrote:
> On Thu, Jul 06, 2017 at 04:48:40PM +0200, Christoph Hellwig wrote:
> 
> > Just did the whole batch (patch below), but it seems like using a
> > __bitwise type in SYSCALL_DEFINE* will always give warnings like:
> > 
> > fs/read_write.c:1095:1: warning: cast to restricted __kernel_rwf_t
> > 
> > which I'm not sure to deal with..
> 
> #define __SC_CAST(t, a) (__force t) a

doesn't seem to make a difference..

Re: [git pull] vfs.git part 3

[Christoph Hellwig]

On Thu, Jul 06, 2017 at 04:03:30PM +0100, Al Viro wrote:
> #define __SC_CAST(t, a) (__force t) a
> 
> in syscalls.h

Hmm.

> 
> > index a2d4a8ac94ca..a04adbc70ddf 100644
> > --- a/include/uapi/linux/aio_abi.h
> > +++ b/include/uapi/linux/aio_abi.h
> > @@ -28,6 +28,7 @@
> >  #define __LINUX__AIO_ABI_H
> >  
> >  #include 
> > +#include 
> 
> Um...  Includes of non-uapi in uapi are wrong.  What do you need
> fs.h for, anyway?  Just put the typedef into uapi/linux/types.h
> and be done with that...

We automatically get the non-uapi one for userspace.  And we do in
fact need to write it that way as it will not show up as uapi/ in
userspace.

But yes, the type could be taken to types.h

> 
> > +#defineBLK_STS_OK 0
> > +#define BLK_STS_NOTSUPP((__force blk_status_t)1)
> > +#define BLK_STS_TIMEOUT((__force blk_status_t)2)
> > +#define BLK_STS_NOSPC  ((__force blk_status_t)3)
> > +#define BLK_STS_TRANSPORT  ((__force blk_status_t)4)
> > +#define BLK_STS_TARGET ((__force blk_status_t)5)
> > +#define BLK_STS_NEXUS  ((__force blk_status_t)6)
> 
> WTF is that doing here?  If nothing else, it's a userland namespace
> pollution; typedefs with such names are OK in the kernel, but not
> is something that might be included from userland.  And that chunk
> doesn't seem to have anything to do with the rest of the patch...

It doesn't  that was the copy and paste of the __bitwise boilerplate
I staeted with..

Re: [git pull] vfs.git part 3

[Christoph Hellwig]

On Thu, Jul 06, 2017 at 04:03:30PM +0100, Al Viro wrote:
> #define __SC_CAST(t, a) (__force t) a
> 
> in syscalls.h

Hmm.

> 
> > index a2d4a8ac94ca..a04adbc70ddf 100644
> > --- a/include/uapi/linux/aio_abi.h
> > +++ b/include/uapi/linux/aio_abi.h
> > @@ -28,6 +28,7 @@
> >  #define __LINUX__AIO_ABI_H
> >  
> >  #include 
> > +#include 
> 
> Um...  Includes of non-uapi in uapi are wrong.  What do you need
> fs.h for, anyway?  Just put the typedef into uapi/linux/types.h
> and be done with that...

We automatically get the non-uapi one for userspace.  And we do in
fact need to write it that way as it will not show up as uapi/ in
userspace.

But yes, the type could be taken to types.h

> 
> > +#defineBLK_STS_OK 0
> > +#define BLK_STS_NOTSUPP((__force blk_status_t)1)
> > +#define BLK_STS_TIMEOUT((__force blk_status_t)2)
> > +#define BLK_STS_NOSPC  ((__force blk_status_t)3)
> > +#define BLK_STS_TRANSPORT  ((__force blk_status_t)4)
> > +#define BLK_STS_TARGET ((__force blk_status_t)5)
> > +#define BLK_STS_NEXUS  ((__force blk_status_t)6)
> 
> WTF is that doing here?  If nothing else, it's a userland namespace
> pollution; typedefs with such names are OK in the kernel, but not
> is something that might be included from userland.  And that chunk
> doesn't seem to have anything to do with the rest of the patch...

It doesn't  that was the copy and paste of the __bitwise boilerplate
I staeted with..

Re: [git pull] vfs.git part 3

[Al Viro]

On Thu, Jul 06, 2017 at 04:48:40PM +0200, Christoph Hellwig wrote:

> Just did the whole batch (patch below), but it seems like using a
> __bitwise type in SYSCALL_DEFINE* will always give warnings like:
> 
> fs/read_write.c:1095:1: warning: cast to restricted __kernel_rwf_t
> 
> which I'm not sure to deal with..

#define __SC_CAST(t, a) (__force t) a

in syscalls.h

> index a2d4a8ac94ca..a04adbc70ddf 100644
> --- a/include/uapi/linux/aio_abi.h
> +++ b/include/uapi/linux/aio_abi.h
> @@ -28,6 +28,7 @@
>  #define __LINUX__AIO_ABI_H
>  
>  #include 
> +#include 

Um...  Includes of non-uapi in uapi are wrong.  What do you need
fs.h for, anyway?  Just put the typedef into uapi/linux/types.h
and be done with that...

> +#define  BLK_STS_OK 0
> +#define BLK_STS_NOTSUPP  ((__force blk_status_t)1)
> +#define BLK_STS_TIMEOUT  ((__force blk_status_t)2)
> +#define BLK_STS_NOSPC((__force blk_status_t)3)
> +#define BLK_STS_TRANSPORT((__force blk_status_t)4)
> +#define BLK_STS_TARGET   ((__force blk_status_t)5)
> +#define BLK_STS_NEXUS((__force blk_status_t)6)

WTF is that doing here?  If nothing else, it's a userland namespace
pollution; typedefs with such names are OK in the kernel, but not
is something that might be included from userland.  And that chunk
doesn't seem to have anything to do with the rest of the patch...

Please, move that on top of current #work.read_write - there's
a fix for buggered vfs_write_iter() in it.

Re: [git pull] vfs.git part 3

[Al Viro]

On Thu, Jul 06, 2017 at 04:48:40PM +0200, Christoph Hellwig wrote:

> Just did the whole batch (patch below), but it seems like using a
> __bitwise type in SYSCALL_DEFINE* will always give warnings like:
> 
> fs/read_write.c:1095:1: warning: cast to restricted __kernel_rwf_t
> 
> which I'm not sure to deal with..

#define __SC_CAST(t, a) (__force t) a

in syscalls.h

> index a2d4a8ac94ca..a04adbc70ddf 100644
> --- a/include/uapi/linux/aio_abi.h
> +++ b/include/uapi/linux/aio_abi.h
> @@ -28,6 +28,7 @@
>  #define __LINUX__AIO_ABI_H
>  
>  #include 
> +#include 

Um...  Includes of non-uapi in uapi are wrong.  What do you need
fs.h for, anyway?  Just put the typedef into uapi/linux/types.h
and be done with that...

> +#define  BLK_STS_OK 0
> +#define BLK_STS_NOTSUPP  ((__force blk_status_t)1)
> +#define BLK_STS_TIMEOUT  ((__force blk_status_t)2)
> +#define BLK_STS_NOSPC((__force blk_status_t)3)
> +#define BLK_STS_TRANSPORT((__force blk_status_t)4)
> +#define BLK_STS_TARGET   ((__force blk_status_t)5)
> +#define BLK_STS_NEXUS((__force blk_status_t)6)

WTF is that doing here?  If nothing else, it's a userland namespace
pollution; typedefs with such names are OK in the kernel, but not
is something that might be included from userland.  And that chunk
doesn't seem to have anything to do with the rest of the patch...

Please, move that on top of current #work.read_write - there's
a fix for buggered vfs_write_iter() in it.

Re: [git pull] vfs.git part 3

[Christoph Hellwig]

On Thu, Jul 06, 2017 at 12:29:12AM +0100, Al Viro wrote:
> On Thu, Jul 06, 2017 at 12:52:35AM +0200, Christoph Hellwig wrote:
> > On Wed, Jul 05, 2017 at 11:38:21PM +0100, Al Viro wrote:
> > > Sure, makes sense - especially since it's not too widely spread yet.
> > 
> > Do you want to do that yourself, or do you want me to look into it?
> 
> I'll do it tomorrow, unless you get to it first...

Just did the whole batch (patch below), but it seems like using a
__bitwise type in SYSCALL_DEFINE* will always give warnings like:

fs/read_write.c:1095:1: warning: cast to restricted __kernel_rwf_t

which I'm not sure to deal with..

--
diff --git a/fs/nfsd/vfs.c b/fs/nfsd/vfs.c
index 38d0383dc7f9..bc69d40c4e8b 100644
--- a/fs/nfsd/vfs.c
+++ b/fs/nfsd/vfs.c
@@ -969,7 +969,7 @@ nfsd_vfs_write(struct svc_rqst *rqstp, struct svc_fh *fhp, 
struct file *file,
int use_wgather;
loff_t  pos = offset;
unsigned intpflags = current->flags;
-   int flags = 0;
+   rwf_t   flags = 0;
 
if (test_bit(RQ_LOCAL, >rq_flags))
/*
diff --git a/fs/read_write.c b/fs/read_write.c
index a2cbc8303dae..bc2db5e5cd19 100644
--- a/fs/read_write.c
+++ b/fs/read_write.c
@@ -633,7 +633,7 @@ unsigned long iov_shorten(struct iovec *iov, unsigned long 
nr_segs, size_t to)
 EXPORT_SYMBOL(iov_shorten);
 
 static ssize_t do_iter_readv_writev(struct file *filp, struct iov_iter *iter,
-   loff_t *ppos, int type, int flags)
+   loff_t *ppos, int type, rwf_t flags)
 {
struct kiocb kiocb;
ssize_t ret;
@@ -655,7 +655,7 @@ static ssize_t do_iter_readv_writev(struct file *filp, 
struct iov_iter *iter,
 
 /* Do it by hand, with file-ops */
 static ssize_t do_loop_readv_writev(struct file *filp, struct iov_iter *iter,
-   loff_t *ppos, int type, int flags)
+   loff_t *ppos, int type, rwf_t flags)
 {
ssize_t ret = 0;
 
@@ -871,7 +871,7 @@ ssize_t compat_rw_copy_check_uvector(int type,
 #endif
 
 static ssize_t do_iter_read(struct file *file, struct iov_iter *iter,
-   loff_t *pos, int flags)
+   loff_t *pos, rwf_t flags)
 {
size_t tot_len;
ssize_t ret = 0;
@@ -899,7 +899,7 @@ static ssize_t do_iter_read(struct file *file, struct 
iov_iter *iter,
 }
 
 ssize_t vfs_iter_read(struct file *file, struct iov_iter *iter, loff_t *ppos,
-   int flags)
+   rwf_t flags)
 {
if (!file->f_op->read_iter)
return -EINVAL;
@@ -908,7 +908,7 @@ ssize_t vfs_iter_read(struct file *file, struct iov_iter 
*iter, loff_t *ppos,
 EXPORT_SYMBOL(vfs_iter_read);
 
 static ssize_t do_iter_write(struct file *file, struct iov_iter *iter,
-   loff_t *pos, int flags)
+   loff_t *pos, rwf_t flags)
 {
size_t tot_len;
ssize_t ret = 0;
@@ -937,7 +937,7 @@ static ssize_t do_iter_write(struct file *file, struct 
iov_iter *iter,
 }
 
 ssize_t vfs_iter_write(struct file *file, struct iov_iter *iter, loff_t *ppos,
-   int flags)
+   rwf_t flags)
 {
if (!file->f_op->write_iter)
return -EINVAL;
@@ -946,7 +946,7 @@ ssize_t vfs_iter_write(struct file *file, struct iov_iter 
*iter, loff_t *ppos,
 EXPORT_SYMBOL(vfs_iter_write);
 
 ssize_t vfs_readv(struct file *file, const struct iovec __user *vec,
- unsigned long vlen, loff_t *pos, int flags)
+ unsigned long vlen, loff_t *pos, rwf_t flags)
 {
struct iovec iovstack[UIO_FASTIOV];
struct iovec *iov = iovstack;
@@ -964,7 +964,7 @@ ssize_t vfs_readv(struct file *file, const struct iovec 
__user *vec,
 EXPORT_SYMBOL(vfs_readv);
 
 ssize_t vfs_writev(struct file *file, const struct iovec __user *vec,
-  unsigned long vlen, loff_t *pos, int flags)
+  unsigned long vlen, loff_t *pos, rwf_t flags)
 {
struct iovec iovstack[UIO_FASTIOV];
struct iovec *iov = iovstack;
@@ -981,7 +981,7 @@ ssize_t vfs_writev(struct file *file, const struct iovec 
__user *vec,
 EXPORT_SYMBOL(vfs_writev);
 
 static ssize_t do_readv(unsigned long fd, const struct iovec __user *vec,
-   unsigned long vlen, int flags)
+   unsigned long vlen, rwf_t flags)
 {
struct fd f = fdget_pos(fd);
ssize_t ret = -EBADF;
@@ -1001,7 +1001,7 @@ static ssize_t do_readv(unsigned long fd, const struct 
iovec __user *vec,
 }
 
 static ssize_t do_writev(unsigned long fd, const struct iovec __user *vec,
-unsigned long vlen, int flags)
+unsigned long vlen, rwf_t flags)
 {
struct fd f = fdget_pos(fd);
ssize_t ret = -EBADF;
@@ -1027,7 +1027,7 @@ static inline loff_t pos_from_hilo(unsigned long high, 
unsigned long low)
 }
 
 static ssize_t do_preadv(unsigned long fd, const struct iovec __user *vec,
-unsigned 

Re: [git pull] vfs.git part 3

[Christoph Hellwig]

On Thu, Jul 06, 2017 at 12:29:12AM +0100, Al Viro wrote:
> On Thu, Jul 06, 2017 at 12:52:35AM +0200, Christoph Hellwig wrote:
> > On Wed, Jul 05, 2017 at 11:38:21PM +0100, Al Viro wrote:
> > > Sure, makes sense - especially since it's not too widely spread yet.
> > 
> > Do you want to do that yourself, or do you want me to look into it?
> 
> I'll do it tomorrow, unless you get to it first...

Just did the whole batch (patch below), but it seems like using a
__bitwise type in SYSCALL_DEFINE* will always give warnings like:

fs/read_write.c:1095:1: warning: cast to restricted __kernel_rwf_t

which I'm not sure to deal with..

--
diff --git a/fs/nfsd/vfs.c b/fs/nfsd/vfs.c
index 38d0383dc7f9..bc69d40c4e8b 100644
--- a/fs/nfsd/vfs.c
+++ b/fs/nfsd/vfs.c
@@ -969,7 +969,7 @@ nfsd_vfs_write(struct svc_rqst *rqstp, struct svc_fh *fhp, 
struct file *file,
int use_wgather;
loff_t  pos = offset;
unsigned intpflags = current->flags;
-   int flags = 0;
+   rwf_t   flags = 0;
 
if (test_bit(RQ_LOCAL, >rq_flags))
/*
diff --git a/fs/read_write.c b/fs/read_write.c
index a2cbc8303dae..bc2db5e5cd19 100644
--- a/fs/read_write.c
+++ b/fs/read_write.c
@@ -633,7 +633,7 @@ unsigned long iov_shorten(struct iovec *iov, unsigned long 
nr_segs, size_t to)
 EXPORT_SYMBOL(iov_shorten);
 
 static ssize_t do_iter_readv_writev(struct file *filp, struct iov_iter *iter,
-   loff_t *ppos, int type, int flags)
+   loff_t *ppos, int type, rwf_t flags)
 {
struct kiocb kiocb;
ssize_t ret;
@@ -655,7 +655,7 @@ static ssize_t do_iter_readv_writev(struct file *filp, 
struct iov_iter *iter,
 
 /* Do it by hand, with file-ops */
 static ssize_t do_loop_readv_writev(struct file *filp, struct iov_iter *iter,
-   loff_t *ppos, int type, int flags)
+   loff_t *ppos, int type, rwf_t flags)
 {
ssize_t ret = 0;
 
@@ -871,7 +871,7 @@ ssize_t compat_rw_copy_check_uvector(int type,
 #endif
 
 static ssize_t do_iter_read(struct file *file, struct iov_iter *iter,
-   loff_t *pos, int flags)
+   loff_t *pos, rwf_t flags)
 {
size_t tot_len;
ssize_t ret = 0;
@@ -899,7 +899,7 @@ static ssize_t do_iter_read(struct file *file, struct 
iov_iter *iter,
 }
 
 ssize_t vfs_iter_read(struct file *file, struct iov_iter *iter, loff_t *ppos,
-   int flags)
+   rwf_t flags)
 {
if (!file->f_op->read_iter)
return -EINVAL;
@@ -908,7 +908,7 @@ ssize_t vfs_iter_read(struct file *file, struct iov_iter 
*iter, loff_t *ppos,
 EXPORT_SYMBOL(vfs_iter_read);
 
 static ssize_t do_iter_write(struct file *file, struct iov_iter *iter,
-   loff_t *pos, int flags)
+   loff_t *pos, rwf_t flags)
 {
size_t tot_len;
ssize_t ret = 0;
@@ -937,7 +937,7 @@ static ssize_t do_iter_write(struct file *file, struct 
iov_iter *iter,
 }
 
 ssize_t vfs_iter_write(struct file *file, struct iov_iter *iter, loff_t *ppos,
-   int flags)
+   rwf_t flags)
 {
if (!file->f_op->write_iter)
return -EINVAL;
@@ -946,7 +946,7 @@ ssize_t vfs_iter_write(struct file *file, struct iov_iter 
*iter, loff_t *ppos,
 EXPORT_SYMBOL(vfs_iter_write);
 
 ssize_t vfs_readv(struct file *file, const struct iovec __user *vec,
- unsigned long vlen, loff_t *pos, int flags)
+ unsigned long vlen, loff_t *pos, rwf_t flags)
 {
struct iovec iovstack[UIO_FASTIOV];
struct iovec *iov = iovstack;
@@ -964,7 +964,7 @@ ssize_t vfs_readv(struct file *file, const struct iovec 
__user *vec,
 EXPORT_SYMBOL(vfs_readv);
 
 ssize_t vfs_writev(struct file *file, const struct iovec __user *vec,
-  unsigned long vlen, loff_t *pos, int flags)
+  unsigned long vlen, loff_t *pos, rwf_t flags)
 {
struct iovec iovstack[UIO_FASTIOV];
struct iovec *iov = iovstack;
@@ -981,7 +981,7 @@ ssize_t vfs_writev(struct file *file, const struct iovec 
__user *vec,
 EXPORT_SYMBOL(vfs_writev);
 
 static ssize_t do_readv(unsigned long fd, const struct iovec __user *vec,
-   unsigned long vlen, int flags)
+   unsigned long vlen, rwf_t flags)
 {
struct fd f = fdget_pos(fd);
ssize_t ret = -EBADF;
@@ -1001,7 +1001,7 @@ static ssize_t do_readv(unsigned long fd, const struct 
iovec __user *vec,
 }
 
 static ssize_t do_writev(unsigned long fd, const struct iovec __user *vec,
-unsigned long vlen, int flags)
+unsigned long vlen, rwf_t flags)
 {
struct fd f = fdget_pos(fd);
ssize_t ret = -EBADF;
@@ -1027,7 +1027,7 @@ static inline loff_t pos_from_hilo(unsigned long high, 
unsigned long low)
 }
 
 static ssize_t do_preadv(unsigned long fd, const struct iovec __user *vec,
-unsigned 

Re: [git pull] vfs.git part 3

[Al Viro]

On Thu, Jul 06, 2017 at 12:52:35AM +0200, Christoph Hellwig wrote:
> On Wed, Jul 05, 2017 at 11:38:21PM +0100, Al Viro wrote:
> > Sure, makes sense - especially since it's not too widely spread yet.
> 
> Do you want to do that yourself, or do you want me to look into it?

I'll do it tomorrow, unless you get to it first...

Re: [git pull] vfs.git part 3

[Al Viro]

On Thu, Jul 06, 2017 at 12:52:35AM +0200, Christoph Hellwig wrote:
> On Wed, Jul 05, 2017 at 11:38:21PM +0100, Al Viro wrote:
> > Sure, makes sense - especially since it's not too widely spread yet.
> 
> Do you want to do that yourself, or do you want me to look into it?

I'll do it tomorrow, unless you get to it first...

Re: [git pull] vfs.git part 3

[Christoph Hellwig]

On Wed, Jul 05, 2017 at 11:38:21PM +0100, Al Viro wrote:
> Sure, makes sense - especially since it's not too widely spread yet.

Do you want to do that yourself, or do you want me to look into it?

Re: [git pull] vfs.git part 3

[Christoph Hellwig]

On Wed, Jul 05, 2017 at 11:38:21PM +0100, Al Viro wrote:
> Sure, makes sense - especially since it's not too widely spread yet.

Do you want to do that yourself, or do you want me to look into it?

Re: [git pull] vfs.git part 3

[Al Viro]

On Wed, Jul 05, 2017 at 02:51:43PM -0700, Linus Torvalds wrote:
> On Wed, Jul 5, 2017 at 12:14 AM, Al Viro  wrote:
> >
> > Christoph's fs/read_write.c series - consolidation and cleanups.
> 
> Side note - when looking through this, it struck me how confusing that
> "int flags" argument was.
> 
> We have a ton of "flags" in the filesystem layer, and how all the
> read/write helpers take them too, and it's really hard to see what
> kind of flags they are.
> 
> Could we perhaps make those RWF_xyz flags have a nice bitwise type,
> and use that type in the argument list, so that not only could there
> be some sparse typechecking, but the functions that pass flags on to
> each other would automatically have a certain amount of actual
> self-documenting prototypes?
> 
> So when you look at one of those vfs_iter_write() or whatever
> functions, you just see *what* flags the flags argument is.
> 
> Because "int flags" really is the worst. It's the wrong type anyway
> (at least make it unsigned if it's a collection of bits), but it's
> also very ambiguous indeed when there are so many other flags that are
> often used/tested in the same functions (there's the "iter" flagsm,
> there's file->f_mode, there's just a lot of different flags going on,
> and the "int flags" is the least well documented of them all,
> particularly since 99.9% of all users just pass in zero).

Sure, makes sense - especially since it's not too widely spread yet.
A side note right back at you - POLL... stuff.  I'd redone the old
"hunt the buggy ->poll() instances down" series (took about 12 hours
total), got it to the point where all remaining sparse warnings about
that type are for genuine bugs.  It goes like that:

define __poll_t, annotate constants
Type is controlled by ifdef - it's unsigned int unless CHECK_POLL is
defined and a bitwise type otherwise.
->poll() methods should return __poll_t
anntotate the places where ->poll() return values go
annotate poll-related wait keys
annotate poll_table_struct ->_key
That ends all infrastructure work.  Methods declarations are annotated,
instances are *not*.  Due to that ifdef CHECK_POLL, normal builds, including
normal sparse builds, are unaffected; with CF=-DCHECK_POLL you get __poll_t
warnings.
cris: annotate ->poll() instances
ia64: annotate ->poll() instances
mips: annotate ->poll() instances
ppc: annotate ->poll() instances
um: annotate ->poll() instances
x86: annotate ->poll() instances
block: annotate ->poll() instances
crypto: annotate ->poll() instances
acpi: annotate ->poll() instances
sound: annotate ->poll() instances
tomoyo: annotate ->poll() instances
net: annotate ->poll() instances
ipc, kernel, mm: annotate ->poll() instances
fs: annotate ->poll() instances
media: annotate ->poll() instances
the rest of drivers/*: annotate ->poll() instances
These can be folded and split as desired - almost up to per-instance.  It's
pretty much "turn unsigned int foo_poll(...) into __poll_t foo_poll(...),
turn unsigned int mask; in it into __poll_t mask;" kind of stuff.  Can go
on per-subsystem basis just fine - again, normal builds are completely 
unaffected.

scif: annotate scif_pollepd
vhost: annotate vhost_poll
dmabuf: annotate dma_buf->active

Several drivers playing games of their own with POLL... bitmaps.

annotate fs/select.c and fs/eventpoll.c
That, of course, can move up right after the infrastructure.


Again, can be reordered in front of the entire queue.  Some are brainos
(POLL_IN instead of POLLIN - compare the kernel definitions of those),
some are "what do you mean, no returning -E... from ->poll()?".  However,
there's the shitty part - poll/epoll ABI mess.  POLLWR... and POLLRDHUP
are architecture-dependent; EPOLL counterparts are not and both are parts
of ABI.  Consider e.g. sparc:
#define POLLWRNORM  POLLOUT [4, that is]
#define POLLWRBAND  256
#define POLLMSG 512
#define POLLREMOVE  1024
#define POLLRDHUP   2048
and compare with
#define EPOLLWRNORM 0x0100
#define EPOLLWRBAND 0x0200
#define EPOLLMSG0x0400
#define EPOLLRDHUP  0x2000

EPOLLRDHUP is never matched.  Neither is EPOLLMSG (nothing raises
POLLREMOVE, but then nothing raises POLLMSG either).  EPOLLWRBAND
is not matched either (that would be POLLMSG).  And EPOLLWRNORM
is matched when we raise POLLWRBAND.

sparc is the worst case in that respect; mips is somewhat better -
there we have
#define POLLWRNORM  POLLOUT
#define POLLWRBAND  0x0100
and everything else is default.  IOW, EPOLLWRBAND is never matched
and EPOLLWRNORM is matched when we raise POLLWRBAND.  Several other
architectures are like mips (m68k and even more exotic stuff).

I'm not sure what to do about that.  Davem is probably in the best
position to tell...

It 

Re: [git pull] vfs.git part 3

[Al Viro]

On Wed, Jul 05, 2017 at 02:51:43PM -0700, Linus Torvalds wrote:
> On Wed, Jul 5, 2017 at 12:14 AM, Al Viro  wrote:
> >
> > Christoph's fs/read_write.c series - consolidation and cleanups.
> 
> Side note - when looking through this, it struck me how confusing that
> "int flags" argument was.
> 
> We have a ton of "flags" in the filesystem layer, and how all the
> read/write helpers take them too, and it's really hard to see what
> kind of flags they are.
> 
> Could we perhaps make those RWF_xyz flags have a nice bitwise type,
> and use that type in the argument list, so that not only could there
> be some sparse typechecking, but the functions that pass flags on to
> each other would automatically have a certain amount of actual
> self-documenting prototypes?
> 
> So when you look at one of those vfs_iter_write() or whatever
> functions, you just see *what* flags the flags argument is.
> 
> Because "int flags" really is the worst. It's the wrong type anyway
> (at least make it unsigned if it's a collection of bits), but it's
> also very ambiguous indeed when there are so many other flags that are
> often used/tested in the same functions (there's the "iter" flagsm,
> there's file->f_mode, there's just a lot of different flags going on,
> and the "int flags" is the least well documented of them all,
> particularly since 99.9% of all users just pass in zero).

Sure, makes sense - especially since it's not too widely spread yet.
A side note right back at you - POLL... stuff.  I'd redone the old
"hunt the buggy ->poll() instances down" series (took about 12 hours
total), got it to the point where all remaining sparse warnings about
that type are for genuine bugs.  It goes like that:

define __poll_t, annotate constants
Type is controlled by ifdef - it's unsigned int unless CHECK_POLL is
defined and a bitwise type otherwise.
->poll() methods should return __poll_t
anntotate the places where ->poll() return values go
annotate poll-related wait keys
annotate poll_table_struct ->_key
That ends all infrastructure work.  Methods declarations are annotated,
instances are *not*.  Due to that ifdef CHECK_POLL, normal builds, including
normal sparse builds, are unaffected; with CF=-DCHECK_POLL you get __poll_t
warnings.
cris: annotate ->poll() instances
ia64: annotate ->poll() instances
mips: annotate ->poll() instances
ppc: annotate ->poll() instances
um: annotate ->poll() instances
x86: annotate ->poll() instances
block: annotate ->poll() instances
crypto: annotate ->poll() instances
acpi: annotate ->poll() instances
sound: annotate ->poll() instances
tomoyo: annotate ->poll() instances
net: annotate ->poll() instances
ipc, kernel, mm: annotate ->poll() instances
fs: annotate ->poll() instances
media: annotate ->poll() instances
the rest of drivers/*: annotate ->poll() instances
These can be folded and split as desired - almost up to per-instance.  It's
pretty much "turn unsigned int foo_poll(...) into __poll_t foo_poll(...),
turn unsigned int mask; in it into __poll_t mask;" kind of stuff.  Can go
on per-subsystem basis just fine - again, normal builds are completely 
unaffected.

scif: annotate scif_pollepd
vhost: annotate vhost_poll
dmabuf: annotate dma_buf->active

Several drivers playing games of their own with POLL... bitmaps.

annotate fs/select.c and fs/eventpoll.c
That, of course, can move up right after the infrastructure.


Again, can be reordered in front of the entire queue.  Some are brainos
(POLL_IN instead of POLLIN - compare the kernel definitions of those),
some are "what do you mean, no returning -E... from ->poll()?".  However,
there's the shitty part - poll/epoll ABI mess.  POLLWR... and POLLRDHUP
are architecture-dependent; EPOLL counterparts are not and both are parts
of ABI.  Consider e.g. sparc:
#define POLLWRNORM  POLLOUT [4, that is]
#define POLLWRBAND  256
#define POLLMSG 512
#define POLLREMOVE  1024
#define POLLRDHUP   2048
and compare with
#define EPOLLWRNORM 0x0100
#define EPOLLWRBAND 0x0200
#define EPOLLMSG0x0400
#define EPOLLRDHUP  0x2000

EPOLLRDHUP is never matched.  Neither is EPOLLMSG (nothing raises
POLLREMOVE, but then nothing raises POLLMSG either).  EPOLLWRBAND
is not matched either (that would be POLLMSG).  And EPOLLWRNORM
is matched when we raise POLLWRBAND.

sparc is the worst case in that respect; mips is somewhat better -
there we have
#define POLLWRNORM  POLLOUT
#define POLLWRBAND  0x0100
and everything else is default.  IOW, EPOLLWRBAND is never matched
and EPOLLWRNORM is matched when we raise POLLWRBAND.  Several other
architectures are like mips (m68k and even more exotic stuff).

I'm not sure what to do about that.  Davem is probably in the best
position to tell...

It might be worth merging the 

Re: [git pull] vfs.git part 3

[Linus Torvalds]

On Wed, Jul 5, 2017 at 12:14 AM, Al Viro  wrote:
>
> Christoph's fs/read_write.c series - consolidation and cleanups.

Side note - when looking through this, it struck me how confusing that
"int flags" argument was.

We have a ton of "flags" in the filesystem layer, and how all the
read/write helpers take them too, and it's really hard to see what
kind of flags they are.

Could we perhaps make those RWF_xyz flags have a nice bitwise type,
and use that type in the argument list, so that not only could there
be some sparse typechecking, but the functions that pass flags on to
each other would automatically have a certain amount of actual
self-documenting prototypes?

So when you look at one of those vfs_iter_write() or whatever
functions, you just see *what* flags the flags argument is.

Because "int flags" really is the worst. It's the wrong type anyway
(at least make it unsigned if it's a collection of bits), but it's
also very ambiguous indeed when there are so many other flags that are
often used/tested in the same functions (there's the "iter" flagsm,
there's file->f_mode, there's just a lot of different flags going on,
and the "int flags" is the least well documented of them all,
particularly since 99.9% of all users just pass in zero).

Hmm?

Linus

Re: [git pull] vfs.git part 3

[Linus Torvalds]

On Wed, Jul 5, 2017 at 12:14 AM, Al Viro  wrote:
>
> Christoph's fs/read_write.c series - consolidation and cleanups.

Side note - when looking through this, it struck me how confusing that
"int flags" argument was.

We have a ton of "flags" in the filesystem layer, and how all the
read/write helpers take them too, and it's really hard to see what
kind of flags they are.

Could we perhaps make those RWF_xyz flags have a nice bitwise type,
and use that type in the argument list, so that not only could there
be some sparse typechecking, but the functions that pass flags on to
each other would automatically have a certain amount of actual
self-documenting prototypes?

So when you look at one of those vfs_iter_write() or whatever
functions, you just see *what* flags the flags argument is.

Because "int flags" really is the worst. It's the wrong type anyway
(at least make it unsigned if it's a collection of bits), but it's
also very ambiguous indeed when there are so many other flags that are
often used/tested in the same functions (there's the "iter" flagsm,
there's file->f_mode, there's just a lot of different flags going on,
and the "int flags" is the least well documented of them all,
particularly since 99.9% of all users just pass in zero).

Hmm?

Linus