Hi наб,
On Sat, Aug 17, 2024 at 12:35:27AM GMT, наб wrote:
> There's no point documenting this syscall at any point in time,
> because it changed constantly. A post-mortem summary I believe to be
> comprehensive is included below.
>
> The #include <sys/kdaemon.h> was removed in glibc 2.23:
> commit eed3e1eb79bcfa9b52609fd875fa2d522e2d6bce
> Author: Joseph Myers <joseph@xxxxxxxxxxxxxxxx>
> Date: Mon Dec 14 22:52:15 2015 +0000
>
> Make obsolete syscall wrappers into compat symbols (bug 18472).
>
> * bdflush: in Linux 2.6, does nothing if present.
>
> [...]
> generated for such aliases. Those five syscalls are then made into
> compat symbols (obsoleted in glibc 2.23, so future ports won't have
> these symbols at all), with the header <sys/kdaemon.h> declaring
> bdflush being removed. When we move to 3.2 as minimum kernel version,
> the same can be done for nfsservctl (removed in Linux 3.1) as well.
> [...]
>
> Appears in 1.1.3
> (func=0 turns the calling process into the bdflush daemon,
> func=1 return sync_old_buffers();,
> func>=2 is parameters):
> Author: Linus Torvalds <torvalds@xxxxxxxxxxxxxxxxxxx>
> Import 1.1.3
>
> +/* This is the interface to bdflush. As we get more sophisticated, we can
> + * pass tuning parameters to this "process", to adjust how it behaves. If you
> + * invoke this again after you have done this once, you would simply modify
> + * the tuning parameters. We would want to verify each parameter, however,
> + * to make sure that it is reasonable. */
> +
> +asmlinkage int sys_bdflush(int func, int data)
> +{
>
> bdflush() is just a kernel thread, and func 0 is just return 0; since 1.3.50:
> Author: Linus Torvalds <torvalds@xxxxxxxxxxxxxxxxxxx>
> Import 1.3.50
>
> - /* Basically func 0 means start, 1 means read param 1, 2 means write param 1, etc */
> + /* Basically func 1 means read param 1, 2 means write param 1, etc */
> if (func >= 2) {
> i = (func-2) >> 1;
> if (i < 0 || i >= N_PARAM)
> @@ -1930,13 +1845,32 @@ asmlinkage int sys_bdflush(int func, long data)
> bdf_prm.data[i] = data;
> return 0;
> };
> +
> + /* Having func 0 used to launch the actual bdflush and then never
> + return (unless explicitly killed). We return zero here to
> + remain semi-compatible with present update(8) programs. */
> +
> + return 0;
> +}
> +
> +/* This is the actual bdflush daemon itself. It used to be started from
> + * the syscall above, but now we launch it ourselves internally with
> + * kernel_thread(...) directly after the first thread in init/main.c */
> +
> +int bdflush(void * unused) {
>
> func 1 is actually exit(0) since 2.3.23pre1:
> Author: Linus Torvalds <torvalds@xxxxxxxxxxxxxxxxxxx>
> Import 2.3.23pre1
>
> if (func == 1) {
> + /* do_exit directly and let kupdate to do its work alone. */
> + do_exit(0);
> +#if 0 /* left here as it's the only example of lazy-mm-stuff used from
> + a syscall that doesn't care about the current mm context. */
>
> fund!=0 is a return 0 since 2.5.12:
> Author: Andrew Morton <akpm@xxxxxxxxxx>
> Date: Mon Apr 29 23:52:10 2002 -0700
>
> [PATCH] writeback from address spaces
>
> [ I reversed the order in which writeback walks the superblock's
> dirty inodes. It sped up dbench's unlink phase greatly. I'm
> such a sleaze ]
>
> The core writeback patch. Switches file writeback from the dirty
> buffer LRU over to address_space.dirty_pages.
>
> - The buffer LRU is removed
>
> - The buffer hash is removed (uses blockdev pagecache lookups)
>
> - The bdflush and kupdate functions are implemented against
> address_spaces, via pdflush.
>
> [...]
>
> Deprecated since 2.5.52:
> Author: Andrew Morton <akpm@xxxxxxxxx>
> Date: Sat Dec 14 03:16:29 2002 -0800
>
> [PATCH] deprecate use of bdflush()
>
> Patch from Robert Love <rml@xxxxxxxxx>
>
> We can never get rid of it if we do not deprecate it - so do so and
> print a stern warning to those who still run bdflush daemons.
>
> Removed outright in 5.15-rc1:
> commit b48c7236b13cb5ef1b5fdf744aa8841df0f7b43a
> Author: Eric W. Biederman <ebiederm@xxxxxxxxxxxx>
> Date: Tue Jun 29 15:11:44 2021 -0500
>
> exit/bdflush: Remove the deprecated bdflush system call
>
> The bdflush system call has been deprecated for a very long time.
> Recently Michael Schmitz tested[1] and found that the last known
> caller of of the bdflush system call is unaffected by it's removal.
>
> Since the code is not needed delete it.
>
> [1] https://lkml.kernel.org/r/36123b5d-daa0-6c2b-f2d4-a942f069fd54@xxxxxxxxx
>
> Signed-off-by: Ahelenia Ziemiańska <nabijaczleweli@xxxxxxxxxxxxxxxxxx>
> ---
> First two applied.
>
> As for the errors: those are equally as varying as the behaviours,
> and a general summary of "some of them" or "all of them at some point"
> is equally as meaningless and useless as the main DESCRIPTION was.
>
> If you're unfortunate enough to care about bdflush(2) (no one is),
> the same applies for the errors (though, if I'm reading
> https://lkml.kernel.org/r/36123b5d-daa0-6c2b-f2d4-a942f069fd54@xxxxxxxxx
> right, update(8) du jour didn't care about them either,
> so much so it still works with the syscall removed).
Makes sense. I've applied the patch. Thanks!
Have a lovely day!
Alex
>
> man/man2/bdflush.2 | 104 +++++++++++---------------------------------
> man/man2/syscalls.2 | 3 +-
> 2 files changed, 28 insertions(+), 79 deletions(-)
>
> diff --git a/man/man2/bdflush.2 b/man/man2/bdflush.2
> index 8627a42..5b5ed16 100644
> --- a/man/man2/bdflush.2
> +++ b/man/man2/bdflush.2
> @@ -12,91 +12,39 @@ .SH SYNOPSIS
> .nf
> .B #include <sys/kdaemon.h>
> .P
> -.BI "[[deprecated]] int bdflush(int " func ", long *" address );
> -.BI "[[deprecated]] int bdflush(int " func ", long " data );
> +.BI "int bdflush(int " func ", long " data );
> .fi
> .SH DESCRIPTION
> -.IR Note :
> -Since Linux 2.6,
> -.\" As noted in changes in the 2.5.12 source
> -this system call is deprecated and does nothing.
> -It is likely to disappear altogether in a future kernel release.
> -Nowadays, the task performed by
> -.BR bdflush ()
> -is handled by the kernel
> -.I pdflush
> -thread.
> +This system call used to turn the calling process into the
> +.I bdflush
> +daemon,
> +or tune it,
> +or flush the "old buffers".
> +It then progressively lost all of that functionality.
> .P
> -.BR bdflush ()
> -starts, flushes, or tunes the buffer-dirty-flush daemon.
> -Only a privileged process (one with the
> -.B CAP_SYS_ADMIN
> -capability) may call
> -.BR bdflush ().
> -.P
> -If
> -.I func
> -is negative or 0, and no daemon has been started, then
> -.BR bdflush ()
> -enters the daemon code and never returns.
> -.P
> -If
> -.I func
> -is 1,
> -some dirty buffers are written to disk.
> -.P
> -If
> -.I func
> -is 2 or more and is even (low bit is 0), then
> -.I address
> -is the address of a long word,
> -and the tuning parameter numbered
> -.RI "(" "func" "\-2)/2"
> -is returned to the caller in that address.
> -.P
> -If
> -.I func
> -is 3 or more and is odd (low bit is 1), then
> -.I data
> -is a long word,
> -and the kernel sets tuning parameter numbered
> -.RI "(" "func" "\-3)/2"
> -to that value.
> -.P
> -The set of parameters, their values, and their valid ranges
> -are defined in the Linux kernel source file
> -.IR fs/buffer.c .
> -.SH RETURN VALUE
> -If
> -.I func
> -is negative or 0 and the daemon successfully starts,
> -.BR bdflush ()
> -never returns.
> -Otherwise, the return value is 0 on success and \-1 on failure, with
> -.I errno
> -set to indicate the error.
> +See
> +.I fs/buffer.c
> +in the kernel version you're interested in to see what it actually does there.
> .SH ERRORS
> -.TP
> -.B EBUSY
> -An attempt was made to enter the daemon code after
> -another process has already entered.
> -.TP
> -.B EFAULT
> -.I address
> -points outside your accessible address space.
> -.TP
> -.B EINVAL
> -An attempt was made to read or write an invalid parameter number,
> -or to write an invalid value to a parameter.
> -.TP
> -.B EPERM
> -Caller does not have the
> -.B CAP_SYS_ADMIN
> -capability.
> +.B ENOSYS
> +(this system call is unimplemented)
> .SH STANDARDS
> Linux.
> .SH HISTORY
> -Since glibc 2.23, glibc no longer supports this obsolete system call.
> +This system call was introduced in Linux 1.1.3,
> +became effectively obsolete in Linux 1.3.50,
> +mostly useless in Linux 2.3.23,
> +entirely useless in Linux 2.5.12,
> +officially deprecated in Linux 2.5.52,
> +and removed outright in Linux 5.15.
> +.P
> +Sometimes, if
> +.I func
> +was even,
> +.I data
> +actually represented a pointer.
> +.P
> +The header and prototype were removed in glibc 2.23.
> .SH SEE ALSO
> .BR sync (1),
> .BR fsync (2),
> diff --git a/man/man2/syscalls.2 b/man/man2/syscalls.2
> index 7a7d6d7..56bbb59 100644
> --- a/man/man2/syscalls.2
> +++ b/man/man2/syscalls.2
> @@ -170,7 +170,8 @@ .SS System call list
> \fBatomic_cmpxchg_32\fP(2) 2.6.34 m68k only
> \fBbdflush\fP(2) 1.2 T{
> Deprecated (does nothing)
> -since 2.6
> +since 2.6,
> +removed in 5.15
> T}
> \fBbind\fP(2) 2.0 T{
> See notes on \fBsocketcall\fP(2)
> --
> 2.39.2
--
<https://www.alejandro-colomar.es/>
Attachment:
signature.asc
Description: PGP signature
