[v4] linux/termio: remove <termio.h> and struct termio

  The <termio.h> interface is absolutely ancient: it was obsoleted by
<termios.h> already in the first version of POSIX (1988) and thus
predates the very first version of Linux. Unfortunately, some constant
macros are used both by <termio.h> and <termios.h>; particularly
problematic is the baud rate constants since the termio interface
*requires* that the baud rate is set via an enumeration as part of
c_cflag.

In preparation of revamping the termios interface to support the
arbitrary baud rate capability that the Linux kernel has supported
since 2008, remove <termio.h> in the hope that no one still uses this
archaic interface.

Note that there is no actual code in glibc to support termio: it is
purely an unabstracted ioctl() interface.

[ v4: fix botched patch ]
[ v3: rebase onto current master ceeffd970c56893885cbf8382ae34b015f177850 ]
[ v2: remove separate struct termio definitions for mips and sparc ]

Signed-off-by: H. Peter Anvin (Intel) <hpa@zytor.com>
---
 NEWS                                               |  5 +++++
 sysdeps/unix/sysv/linux/Makefile                   |  6 ------
 sysdeps/unix/sysv/linux/bits/ioctl-types.h         | 11 -----------
 sysdeps/unix/sysv/linux/mips/bits/ioctl-types.h    | 12 ------------
 sysdeps/unix/sysv/linux/powerpc/bits/ioctl-types.h | 11 -----------
 sysdeps/unix/sysv/linux/termio.h                   |  6 ------
 6 files changed, 5 insertions(+), 46 deletions(-)
 delete mode 100644 sysdeps/unix/sysv/linux/termio.h
  

On 4/15/25 15:57, H. Peter Anvin wrote:
> The <termio.h> interface is absolutely ancient: it was obsoleted by
> <termios.h> already in the first version of POSIX (1988) and thus
> predates the very first version of Linux. Unfortunately, some constant
> macros are used both by <termio.h> and <termios.h>; particularly
> problematic is the baud rate constants since the termio interface
> *requires* that the baud rate is set via an enumeration as part of
> c_cflag.
> 
> In preparation of revamping the termios interface to support the
> arbitrary baud rate capability that the Linux kernel has supported
> since 2008, remove <termio.h> in the hope that no one still uses this
> archaic interface.
> 
> Note that there is no actual code in glibc to support termio: it is
> purely an unabstracted ioctl() interface.
> 
> [ v4: fix botched patch ]
> [ v3: rebase onto current master ceeffd970c56893885cbf8382ae34b015f177850 ]
> [ v2: remove separate struct termio definitions for mips and sparc ]

Ping?

	-hpa
  

* H. Peter Anvin:

> The <termio.h> interface is absolutely ancient: it was obsoleted by
> <termios.h> already in the first version of POSIX (1988) and thus
> predates the very first version of Linux. Unfortunately, some constant
> macros are used both by <termio.h> and <termios.h>; particularly
> problematic is the baud rate constants since the termio interface
> *requires* that the baud rate is set via an enumeration as part of
> c_cflag.
>
> In preparation of revamping the termios interface to support the
> arbitrary baud rate capability that the Linux kernel has supported
> since 2008, remove <termio.h> in the hope that no one still uses this
> archaic interface.
>
> Note that there is no actual code in glibc to support termio: it is
> purely an unabstracted ioctl() interface.
>
> [ v4: fix botched patch ]
> [ v3: rebase onto current master ceeffd970c56893885cbf8382ae34b015f177850 ]
> [ v2: remove separate struct termio definitions for mips and sparc ]
>
> Signed-off-by: H. Peter Anvin (Intel) <hpa@zytor.com>

Reviewed-by: Florian Weimer <fweimer@redhat.com>

I'm going to apply this based on the previous discussion.

Thanks,
Florian
  

* H. Peter Anvin:

> The <termio.h> interface is absolutely ancient: it was obsoleted by
> <termios.h> already in the first version of POSIX (1988) and thus
> predates the very first version of Linux. Unfortunately, some constant
> macros are used both by <termio.h> and <termios.h>; particularly
> problematic is the baud rate constants since the termio interface
> *requires* that the baud rate is set via an enumeration as part of
> c_cflag.
>
> In preparation of revamping the termios interface to support the
> arbitrary baud rate capability that the Linux kernel has supported
> since 2008, remove <termio.h> in the hope that no one still uses this
> archaic interface.
>
> Note that there is no actual code in glibc to support termio: it is
> purely an unabstracted ioctl() interface.
>
> [ v4: fix botched patch ]
> [ v3: rebase onto current master ceeffd970c56893885cbf8382ae34b015f177850 ]
> [ v2: remove separate struct termio definitions for mips and sparc ]
>
> Signed-off-by: H. Peter Anvin (Intel) <hpa@zytor.com>
> ---
>  NEWS                                               |  5 +++++
>  sysdeps/unix/sysv/linux/Makefile                   |  6 ------
>  sysdeps/unix/sysv/linux/bits/ioctl-types.h         | 11 -----------
>  sysdeps/unix/sysv/linux/mips/bits/ioctl-types.h    | 12 ------------
>  sysdeps/unix/sysv/linux/powerpc/bits/ioctl-types.h | 11 -----------
>  sysdeps/unix/sysv/linux/termio.h                   |  6 ------
>  6 files changed, 5 insertions(+), 46 deletions(-)
>  delete mode 100644 sysdeps/unix/sysv/linux/termio.h

I've pushed that.  It's headed towards Fedora rawhide, so we'll see if
there's any breakage soon.  Likewise for OpenSUSE tumbleweed.

Thanks,
Florian
  

On 4/25/25 08:44, Florian Weimer wrote:
> * H. Peter Anvin:
> 
>> The <termio.h> interface is absolutely ancient: it was obsoleted by
>> <termios.h> already in the first version of POSIX (1988) and thus
>> predates the very first version of Linux. Unfortunately, some constant
>> macros are used both by <termio.h> and <termios.h>; particularly
>> problematic is the baud rate constants since the termio interface
>> *requires* that the baud rate is set via an enumeration as part of
>> c_cflag.
>>
>> In preparation of revamping the termios interface to support the
>> arbitrary baud rate capability that the Linux kernel has supported
>> since 2008, remove <termio.h> in the hope that no one still uses this
>> archaic interface.
>>
>> Note that there is no actual code in glibc to support termio: it is
>> purely an unabstracted ioctl() interface.
>>
>> [ v4: fix botched patch ]
>> [ v3: rebase onto current master ceeffd970c56893885cbf8382ae34b015f177850 ]
>> [ v2: remove separate struct termio definitions for mips and sparc ]
>>
>> Signed-off-by: H. Peter Anvin (Intel) <hpa@zytor.com>
>> ---
>>   NEWS                                               |  5 +++++
>>   sysdeps/unix/sysv/linux/Makefile                   |  6 ------
>>   sysdeps/unix/sysv/linux/bits/ioctl-types.h         | 11 -----------
>>   sysdeps/unix/sysv/linux/mips/bits/ioctl-types.h    | 12 ------------
>>   sysdeps/unix/sysv/linux/powerpc/bits/ioctl-types.h | 11 -----------
>>   sysdeps/unix/sysv/linux/termio.h                   |  6 ------
>>   6 files changed, 5 insertions(+), 46 deletions(-)
>>   delete mode 100644 sysdeps/unix/sysv/linux/termio.h
> 
> I've pushed that.  It's headed towards Fedora rawhide, so we'll see if
> there's any breakage soon.  Likewise for OpenSUSE tumbleweed.
> 

So Linaro reports that this breaks gcc libsanitizer. This is simply 
because of the following lines

sanitizer_common/sanitizer_platform_limits_posix.cpp [488-490]:

#if SANITIZER_GLIBC || SANITIZER_ANDROID
   unsigned struct_termio_sz = sizeof(struct termio);
#endif

struct_termio_sz is then used in sanitizer_common_interceptors_ioctl.inc.

I see a few ways of fixing this:

1. In glibc: remove <termio.h> but leave the definition of struct termio 
in <sys/ioctl.h>

2. Completely remove the above statement in gcc;

3. Do a weird hack in gcc hard-coding the size of legacy struct termio;

4. In gcc, get struct termio from <linux/termios.h> instead. This is 
arguably More Correct anyway; libsanitizer is formally incorrect because 
it gets the termios ioctls wrong (confusing glibc struct termios with 
the kernel one.)

The real issue I guess is if we want to leave the struct termio in 
sys/ioctl.h for the time being to avoid breaking the gcc build, or if we 
want to push a patch into gcc and accept the breakage of libsanitizer in 
the short term. I suspect the answer is no, and that we should thus fall 
back on only removing <termio.h> for now.
  

* H. Peter Anvin:

> On 4/25/25 08:44, Florian Weimer wrote:
>> * H. Peter Anvin:
>> 
>>> The <termio.h> interface is absolutely ancient: it was obsoleted by
>>> <termios.h> already in the first version of POSIX (1988) and thus
>>> predates the very first version of Linux. Unfortunately, some constant
>>> macros are used both by <termio.h> and <termios.h>; particularly
>>> problematic is the baud rate constants since the termio interface
>>> *requires* that the baud rate is set via an enumeration as part of
>>> c_cflag.
>>>
>>> In preparation of revamping the termios interface to support the
>>> arbitrary baud rate capability that the Linux kernel has supported
>>> since 2008, remove <termio.h> in the hope that no one still uses this
>>> archaic interface.
>>>
>>> Note that there is no actual code in glibc to support termio: it is
>>> purely an unabstracted ioctl() interface.
>>>
>>> [ v4: fix botched patch ]
>>> [ v3: rebase onto current master ceeffd970c56893885cbf8382ae34b015f177850 ]
>>> [ v2: remove separate struct termio definitions for mips and sparc ]
>>>
>>> Signed-off-by: H. Peter Anvin (Intel) <hpa@zytor.com>
>>> ---
>>>   NEWS                                               |  5 +++++
>>>   sysdeps/unix/sysv/linux/Makefile                   |  6 ------
>>>   sysdeps/unix/sysv/linux/bits/ioctl-types.h         | 11 -----------
>>>   sysdeps/unix/sysv/linux/mips/bits/ioctl-types.h    | 12 ------------
>>>   sysdeps/unix/sysv/linux/powerpc/bits/ioctl-types.h | 11 -----------
>>>   sysdeps/unix/sysv/linux/termio.h                   |  6 ------
>>>   6 files changed, 5 insertions(+), 46 deletions(-)
>>>   delete mode 100644 sysdeps/unix/sysv/linux/termio.h
>> 
>> I've pushed that.  It's headed towards Fedora rawhide, so we'll see if
>> there's any breakage soon.  Likewise for OpenSUSE tumbleweed.
>> 
>
> So Linaro reports that this breaks gcc libsanitizer. This is simply 
> because of the following lines
>
> sanitizer_common/sanitizer_platform_limits_posix.cpp [488-490]:
>
> #if SANITIZER_GLIBC || SANITIZER_ANDROID
>    unsigned struct_termio_sz = sizeof(struct termio);
> #endif
>
> struct_termio_sz is then used in sanitizer_common_interceptors_ioctl.inc.

There's an upstream/LLVM PR for this:

  [sanitizer_common] Remove interceptors for deprecated struct termio
  <https://github.com/llvm/llvm-project/pull/137403>

I plan to submit a GCC patch once the LLVM change has been merged.
  

* Florian Weimer:

> * H. Peter Anvin:
>
>> On 4/25/25 08:44, Florian Weimer wrote:
>>> * H. Peter Anvin:
>>> 
>>>> The <termio.h> interface is absolutely ancient: it was obsoleted by
>>>> <termios.h> already in the first version of POSIX (1988) and thus
>>>> predates the very first version of Linux. Unfortunately, some constant
>>>> macros are used both by <termio.h> and <termios.h>; particularly
>>>> problematic is the baud rate constants since the termio interface
>>>> *requires* that the baud rate is set via an enumeration as part of
>>>> c_cflag.
>>>>
>>>> In preparation of revamping the termios interface to support the
>>>> arbitrary baud rate capability that the Linux kernel has supported
>>>> since 2008, remove <termio.h> in the hope that no one still uses this
>>>> archaic interface.
>>>>
>>>> Note that there is no actual code in glibc to support termio: it is
>>>> purely an unabstracted ioctl() interface.
>>>>
>>>> [ v4: fix botched patch ]
>>>> [ v3: rebase onto current master ceeffd970c56893885cbf8382ae34b015f177850 ]
>>>> [ v2: remove separate struct termio definitions for mips and sparc ]
>>>>
>>>> Signed-off-by: H. Peter Anvin (Intel) <hpa@zytor.com>
>>>> ---
>>>>   NEWS                                               |  5 +++++
>>>>   sysdeps/unix/sysv/linux/Makefile                   |  6 ------
>>>>   sysdeps/unix/sysv/linux/bits/ioctl-types.h         | 11 -----------
>>>>   sysdeps/unix/sysv/linux/mips/bits/ioctl-types.h    | 12 ------------
>>>>   sysdeps/unix/sysv/linux/powerpc/bits/ioctl-types.h | 11 -----------
>>>>   sysdeps/unix/sysv/linux/termio.h                   |  6 ------
>>>>   6 files changed, 5 insertions(+), 46 deletions(-)
>>>>   delete mode 100644 sysdeps/unix/sysv/linux/termio.h
>>> 
>>> I've pushed that.  It's headed towards Fedora rawhide, so we'll see if
>>> there's any breakage soon.  Likewise for OpenSUSE tumbleweed.
>>> 
>>
>> So Linaro reports that this breaks gcc libsanitizer. This is simply 
>> because of the following lines
>>
>> sanitizer_common/sanitizer_platform_limits_posix.cpp [488-490]:
>>
>> #if SANITIZER_GLIBC || SANITIZER_ANDROID
>>    unsigned struct_termio_sz = sizeof(struct termio);
>> #endif
>>
>> struct_termio_sz is then used in sanitizer_common_interceptors_ioctl.inc.
>
> There's an upstream/LLVM PR for this:
>
>   [sanitizer_common] Remove interceptors for deprecated struct termio
>   <https://github.com/llvm/llvm-project/pull/137403>
>
> I plan to submit a GCC patch once the LLVM change has been merged.

And Tom has a question:

> This isn't quite enough to fix the build. There are also uses of
> TCGETA, TCSETA, TCSETAF, and TCSETAW macros that reference struct
> termio. I'm not sure if these should be removed or if asm.h needs to
> be updated to use struct termios instead.

<https://github.com/llvm/llvm-project/pull/137403#issuecomment-2832595303>

I assume TCGETA etc. are only used with the legacy interface?

Peter, maybe you could comment on the PR?
  

On Apr 27 2025, Florian Weimer wrote:

> <https://github.com/llvm/llvm-project/pull/137403#issuecomment-2832595303>
>
> I assume TCGETA etc. are only used with the legacy interface?

It's still broken:

In file included from /usr/include/asm/ioctl.h:12,
                 from /usr/include/asm/ioctls.h:5,
                 from /usr/include/bits/ioctls.h:23,
                 from /usr/include/sys/ioctl.h:26,
                 from /usr/include/sys/mount.h:28,
                 from ../../../../libsanitizer/sanitizer_common/sanitizer_platform_limits_posix.cpp:64:
../../../../libsanitizer/sanitizer_common/sanitizer_platform_limits_posix.cpp:765:27: error: invalid application of ‘sizeof’ to incomplete type ‘__sanitizer::termio’
  765 |   unsigned IOCTL_TCGETA = TCGETA;
      |                           ^~~~~~
../../../../libsanitizer/sanitizer_common/sanitizer_platform_limits_posix.cpp:769:27: error: invalid application of ‘sizeof’ to incomplete type ‘__sanitizer::termio’
  769 |   unsigned IOCTL_TCSETA = TCSETA;
      |                           ^~~~~~
../../../../libsanitizer/sanitizer_common/sanitizer_platform_limits_posix.cpp:770:28: error: invalid application of ‘sizeof’ to incomplete type ‘__sanitizer::termio’
  770 |   unsigned IOCTL_TCSETAF = TCSETAF;
      |                            ^~~~~~~
../../../../libsanitizer/sanitizer_common/sanitizer_platform_limits_posix.cpp:771:28: error: invalid application of ‘sizeof’ to incomplete type ‘__sanitizer::termio’
  771 |   unsigned IOCTL_TCSETAW = TCSETAW;
      |                            ^~~~~~~
make[4]: *** [Makefile:627: sanitizer_platform_limits_posix.lo] Error 1
make[4]: Leaving directory '/home/abuild/rpmbuild/BUILD/gcc16-testresults-16.0.0+git433-build/gcc-16.0.0+git433/obj-powerpc64le-suse-linux/powerpc64le-suse-linux/libsanitizer/sanitizer_common'
  

This is https://github.com/llvm/llvm-project/pull/138822 now.
  

On May 7, 2025 1:01:40 AM PDT, Andreas Schwab <schwab@suse.de> wrote:
>On Apr 27 2025, Florian Weimer wrote:
>
>> <https://github.com/llvm/llvm-project/pull/137403#issuecomment-2832595303>
>>
>> I assume TCGETA etc. are only used with the legacy interface?
>
>It's still broken:
>
>In file included from /usr/include/asm/ioctl.h:12,
>                 from /usr/include/asm/ioctls.h:5,
>                 from /usr/include/bits/ioctls.h:23,
>                 from /usr/include/sys/ioctl.h:26,
>                 from /usr/include/sys/mount.h:28,
>                 from ../../../../libsanitizer/sanitizer_common/sanitizer_platform_limits_posix.cpp:64:
>../../../../libsanitizer/sanitizer_common/sanitizer_platform_limits_posix.cpp:765:27: error: invalid application of ‘sizeof’ to incomplete type ‘__sanitizer::termio’
>  765 |   unsigned IOCTL_TCGETA = TCGETA;
>      |                           ^~~~~~
>../../../../libsanitizer/sanitizer_common/sanitizer_platform_limits_posix.cpp:769:27: error: invalid application of ‘sizeof’ to incomplete type ‘__sanitizer::termio’
>  769 |   unsigned IOCTL_TCSETA = TCSETA;
>      |                           ^~~~~~
>../../../../libsanitizer/sanitizer_common/sanitizer_platform_limits_posix.cpp:770:28: error: invalid application of ‘sizeof’ to incomplete type ‘__sanitizer::termio’
>  770 |   unsigned IOCTL_TCSETAF = TCSETAF;
>      |                            ^~~~~~~
>../../../../libsanitizer/sanitizer_common/sanitizer_platform_limits_posix.cpp:771:28: error: invalid application of ‘sizeof’ to incomplete type ‘__sanitizer::termio’
>  771 |   unsigned IOCTL_TCSETAW = TCSETAW;
>      |                            ^~~~~~~
>make[4]: *** [Makefile:627: sanitizer_platform_limits_posix.lo] Error 1
>make[4]: Leaving directory '/home/abuild/rpmbuild/BUILD/gcc16-testresults-16.0.0+git433-build/gcc-16.0.0+git433/obj-powerpc64le-suse-linux/powerpc64le-suse-linux/libsanitizer/sanitizer_common'
>

Yes, I believe my comment on the bug report is the right way to fix it.

ioctl is a kernel interface, and the types should be picked up from the kernel headers.
  

On 5/7/25 01:24, Andreas Schwab wrote:
> This is https://github.com/llvm/llvm-project/pull/138822 now.

I don't have any direct objection, but please see my note:

https://github.com/llvm/llvm-project/issues/137321#issuecomment-2836993353