Skip to contentSkip to conversation

tty: Remove dead termiox code

PATCH In review Jann Horn 5 participants 12 messages Dec 3, 2020 – Dec 8, 2020 TTY layer and serial drivers

lore
Share & tools
lore

Jann Horn

author
#0 lore raw

PATCH 1/1 tty: Remove dead termiox code

set_termiox() and the TCGETX handler bail out with -EINVAL immediately if ->termiox is NULL, but there are no code paths that can set ->termiox to a non-NULL pointer; and no such code paths seem to have existed since the termiox mechanism was introduced back in commit 1d65b4a088de ("tty: Add termiox") in v2.6.28. Similarly, no driver actually implements .set_termiox; and it looks like no driver ever has.

Delete this dead code; but leave the definition of struct termiox in the UAPI headers intact.

---
 drivers/tty/tty_ioctl.c    | 61 ++------------------------------------
 include/linux/tty.h        |  1 -
 include/linux/tty_driver.h |  9 ------
 3 files changed, 2 insertions(+), 69 deletions(-)
drivers/tty/tty_ioctl.c +2 -59
index e18f318586ab..4de1c6ddb8ff 100644
@@ -443,51 +443,6 @@ static int get_termio(struct tty_struct *tty, struct termio __user *termio)
 	return 0;
 }
 
-
-#ifdef TCGETX
-
-/**
- *	set_termiox	-	set termiox fields if possible
- *	@tty: terminal
- *	@arg: termiox structure from user
- *	@opt: option flags for ioctl type
- *
- *	Implement the device calling points for the SYS5 termiox ioctl
- *	interface in Linux
- */
-
-static int set_termiox(struct tty_struct *tty, void __user *arg, int opt)
-{
-	struct termiox tnew;
-	struct tty_ldisc *ld;
-
-	if (tty->termiox == NULL)
-		return -EINVAL;
-	if (copy_from_user(&tnew, arg, sizeof(struct termiox)))
-		return -EFAULT;
-
-	ld = tty_ldisc_ref(tty);
-	if (ld != NULL) {
-		if ((opt & TERMIOS_FLUSH) && ld->ops->flush_buffer)
-			ld->ops->flush_buffer(tty);
-		tty_ldisc_deref(ld);
-	}
-	if (opt & TERMIOS_WAIT) {
-		tty_wait_until_sent(tty, 0);
-		if (signal_pending(current))
-			return -ERESTARTSYS;
-	}
-
-	down_write(&tty->termios_rwsem);
-	if (tty->ops->set_termiox)
-		tty->ops->set_termiox(tty, &tnew);
-	up_write(&tty->termios_rwsem);
-	return 0;
-}
-
-#endif
-
-
 #ifdef TIOCGETP
 /*
  * These are deprecated, but there is limited support..
@@ -815,23 +770,11 @@ int tty_mode_ioctl(struct tty_struct *tty, struct file *file,
 		return ret;
 #endif
 #ifdef TCGETX
-	case TCGETX: {
-		struct termiox ktermx;
-		if (real_tty->termiox == NULL)
-			return -EINVAL;
-		down_read(&real_tty->termios_rwsem);
-		memcpy(&ktermx, real_tty->termiox, sizeof(struct termiox));
-		up_read(&real_tty->termios_rwsem);
-		if (copy_to_user(p, &ktermx, sizeof(struct termiox)))
-			ret = -EFAULT;
-		return ret;
-	}
+	case TCGETX:
 	case TCSETX:
-		return set_termiox(real_tty, p, 0);
 	case TCSETXW:
-		return set_termiox(real_tty, p, TERMIOS_WAIT);
 	case TCSETXF:
-		return set_termiox(real_tty, p, TERMIOS_FLUSH);
+		return -EINVAL;
 #endif		
 	case TIOCGSOFTCAR:
 		copy_termios(real_tty, &kterm);
include/linux/tty.h +0 -1
index a99e9b8e4e31..52f5544bcd85 100644
@@ -303,7 +303,6 @@ struct tty_struct {
 	spinlock_t flow_lock;
 	/* Termios values are protected by the termios rwsem */
 	struct ktermios termios, termios_locked;
-	struct termiox *termiox;	/* May be NULL for unsupported */
 	char name[64];
 	struct pid *pgrp;		/* Protected by ctrl lock */
 	struct pid *session;
include/linux/tty_driver.h +0 -9
index 358446247ccd..61c3372d3f32 100644
@@ -224,14 +224,6 @@
  *	line). See tty_do_resize() if you need to wrap the standard method
  *	in your own logic - the usual case.
  *
- * void (*set_termiox)(struct tty_struct *tty, struct termiox *new);
- *
- *	Called when the device receives a termiox based ioctl. Passes down
- *	the requested data from user space. This method will not be invoked
- *	unless the tty also has a valid tty->termiox pointer.
- *
- *	Optional: Called under the termios lock
- *
  * int (*get_icount)(struct tty_struct *tty, struct serial_icounter *icount);
  *
  *	Called when the device receives a TIOCGICOUNT ioctl. Passed a kernel
@@ -285,7 +277,6 @@ struct tty_operations {
 	int (*tiocmset)(struct tty_struct *tty,
 			unsigned int set, unsigned int clear);
 	int (*resize)(struct tty_struct *tty, struct winsize *ws);
-	int (*set_termiox)(struct tty_struct *tty, struct termiox *tnew);
 	int (*get_icount)(struct tty_struct *tty,
 				struct serial_icounter_struct *icount);
 	int  (*get_serial)(struct tty_struct *tty, struct serial_struct *p);

base-commit: 3bb61aa61828499a7d0f5e560051625fd02ae7e4

Signature · 1 line
2.29.2.576.ga3fc446d84-goog
#1 lore raw
Quoted 10 lines · Jann Horn

set_termiox() and the TCGETX handler bail out with -EINVAL immediately if ->termiox is NULL, but there are no code paths that can set ->termiox to a non-NULL pointer; and no such code paths seem to have existed since the termiox mechanism was introduced back in commit 1d65b4a088de ("tty: Add termiox") in v2.6.28. Similarly, no driver actually implements .set_termiox; and it looks like no driver ever has.

Delete this dead code; but leave the definition of struct termiox in the UAPI headers intact.

Crazy that no one uses it. Nice cleanup, thanks for doing this.

greg k-h

maintainer
#2 lore raw
In reply to Jann Horn
Quoted 7 lines · Jann Horn

set_termiox() and the TCGETX handler bail out with -EINVAL immediately if ->termiox is NULL, but there are no code paths that can set ->termiox to a non-NULL pointer; and no such code paths seem to have existed since the termiox mechanism was introduced back in commit 1d65b4a088de ("tty: Add termiox") in v2.6.28. Similarly, no driver actually implements .set_termiox; and it looks like no driver ever has.

Nice!

Quoted 2 lines

Delete this dead code; but leave the definition of struct termiox in the UAPI headers intact.

I am thinking -- can/should we mark the structure as deprecated so that userspace stops using it eventually?

Signature · 1 line
js
#3 lore raw
Quoted 9 lines · Jiri Slaby
Quoted 7 lines · Jann Horn

set_termiox() and the TCGETX handler bail out with -EINVAL immediately if ->termiox is NULL, but there are no code paths that can set ->termiox to a non-NULL pointer; and no such code paths seem to have existed since the termiox mechanism was introduced back in commit 1d65b4a088de ("tty: Add termiox") in v2.6.28. Similarly, no driver actually implements .set_termiox; and it looks like no driver ever has.

Quoted 7 lines

Nice!

Quoted 2 lines

Delete this dead code; but leave the definition of struct termiox in the UAPI headers intact.

I am thinking -- can/should we mark the structure as deprecated so that userspace stops using it eventually?

If it doesn't do anything, how can userspace even use it today? :)

maintainer
#4 lore raw
Quoted 12 lines · Greg Kroah-Hartman
Quoted 11 lines · Jiri Slaby
Quoted 7 lines · Jann Horn

set_termiox() and the TCGETX handler bail out with -EINVAL immediately if ->termiox is NULL, but there are no code paths that can set ->termiox to a non-NULL pointer; and no such code paths seem to have existed since the termiox mechanism was introduced back in commit 1d65b4a088de ("tty: Add termiox") in v2.6.28. Similarly, no driver actually implements .set_termiox; and it looks like no driver ever has.

Nice!

Quoted 7 lines
Quoted 5 lines
Quoted 2 lines

Delete this dead code; but leave the definition of struct termiox in the UAPI headers intact.

I am thinking -- can/should we mark the structure as deprecated so that userspace stops using it eventually?

If it doesn't do anything, how can userspace even use it today? :)

Well, right. I am in favor to remove it, BUT: what if someone tries that ioctl and bails out if EINVAL is returned. I mean: if they define a local var of that struct type and pass it to the ioctl, we would break the build by removing the struct completely. Even if the code didn't do anything useful, it still could be built. So is this very potential breakage OK?

thanks,

Signature · 1 line
js
#5 lore raw
Quoted 19 lines · Jiri Slaby
Quoted 18 lines · Greg Kroah-Hartman
Quoted 16 lines · Jiri Slaby
Quoted 7 lines · Jann Horn

set_termiox() and the TCGETX handler bail out with -EINVAL immediately if ->termiox is NULL, but there are no code paths that can set ->termiox to a non-NULL pointer; and no such code paths seem to have existed since the termiox mechanism was introduced back in commit 1d65b4a088de ("tty: Add termiox") in v2.6.28. Similarly, no driver actually implements .set_termiox; and it looks like no driver ever has.

Nice!

Quoted 2 lines

Delete this dead code; but leave the definition of struct termiox in the UAPI headers intact.

I am thinking -- can/should we mark the structure as deprecated so that userspace stops using it eventually?

Quoted 7 lines
Quoted 1 line

If it doesn't do anything, how can userspace even use it today? :)

Well, right. I am in favor to remove it, BUT: what if someone tries that ioctl and bails out if EINVAL is returned. I mean: if they define a local var of that struct type and pass it to the ioctl, we would break the build by removing the struct completely. Even if the code didn't do anything useful, it still could be built. So is this very potential breakage OK?

I'm sorry, but I don't understand. This is a kernel-internal-only structure, right? If someone today tries to call these ioctls, they will get a -EINVAL error as no serial driver in the tree supports them.

If we remove the structure (i.e. what this patch does), and someone makes an ioctl call, they will still get the same -EINVAL error they did before.

So nothing has changed as far as userspace can tell.

Now if they have an out-of-tree serial driver that does implement this call, then yes, they will have problems, but that's not our problem, that is theirs for not ever submitting their code. We don't support in-kernel apis with no in-kernel users.

Or am I still confused?

thanks,

greg k-h

maintainer
#6 lore raw
Quoted 12 lines · Greg Kroah-Hartman
Quoted 11 lines · Jiri Slaby
Quoted 10 lines · Greg Kroah-Hartman
Quoted 9 lines · Jiri Slaby
Quoted 7 lines · Jann Horn

set_termiox() and the TCGETX handler bail out with -EINVAL immediately if ->termiox is NULL, but there are no code paths that can set ->termiox to a non-NULL pointer; and no such code paths seem to have existed since the termiox mechanism was introduced back in commit 1d65b4a088de ("tty: Add termiox") in v2.6.28. Similarly, no driver actually implements .set_termiox; and it looks like no driver ever has.

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

Nice!

Quoted 2 lines

Delete this dead code; but leave the definition of struct termiox in the UAPI headers intact.

Note this ^^^^^. He is talking about _not_ touching the definition in the UAPI header. Does the rest below makes more sense now?

Quoted 32 lines
Quoted 10 lines
Quoted 4 lines
Quoted 2 lines

I am thinking -- can/should we mark the structure as deprecated so that userspace stops using it eventually?

If it doesn't do anything, how can userspace even use it today? :)

Well, right. I am in favor to remove it, BUT: what if someone tries that ioctl and bails out if EINVAL is returned. I mean: if they define a local var of that struct type and pass it to the ioctl, we would break the build by removing the struct completely. Even if the code didn't do anything useful, it still could be built. So is this very potential breakage OK?

I'm sorry, but I don't understand. This is a kernel-internal-only structure, right? If someone today tries to call these ioctls, they will get a -EINVAL error as no serial driver in the tree supports them.

If we remove the structure (i.e. what this patch does), and someone makes an ioctl call, they will still get the same -EINVAL error they did before.

So nothing has changed as far as userspace can tell.

Now if they have an out-of-tree serial driver that does implement this call, then yes, they will have problems, but that's not our problem, that is theirs for not ever submitting their code. We don't support in-kernel apis with no in-kernel users.

Or am I still confused?

thanks,

greg k-h

Signature · 1 line
js
#7 lore raw
Quoted 13 lines · Jiri Slaby
Quoted 12 lines · Greg Kroah-Hartman
Quoted 11 lines · Jiri Slaby
Quoted 10 lines · Greg Kroah-Hartman
Quoted 9 lines · Jiri Slaby
Quoted 7 lines · Jann Horn

set_termiox() and the TCGETX handler bail out with -EINVAL immediately if ->termiox is NULL, but there are no code paths that can set ->termiox to a non-NULL pointer; and no such code paths seem to have existed since the termiox mechanism was introduced back in commit 1d65b4a088de ("tty: Add termiox") in v2.6.28. Similarly, no driver actually implements .set_termiox; and it looks like no driver ever has.

Quoted 7 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

Nice!

Quoted 2 lines

Delete this dead code; but leave the definition of struct termiox in the UAPI headers intact.

Note this ^^^^^. He is talking about _not_ touching the definition in the UAPI header. Does the rest below makes more sense now?

No, I'm still confused :)

We can't touch the UAPI definitions, but the fact that this api never did anything still is ok as after this patch it continues to not do anything.

I'm confused as to what you are proposing...

greg k-h

#8 lore raw
Quoted 5 lines · Greg Kroah-Hartman
Quoted 4 lines · Jiri Slaby
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines · Jiri Slaby
Quoted 3 lines · Jann Horn
Quoted 2 lines

Delete this dead code; but leave the definition of struct termiox in the UAPI headers intact.

[was snipped]

Quoted 6 lines
Quoted 5 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

I am thinking -- can/should we mark the structure as deprecated so that userspace stops using it eventually?

Note this ^^^^^. He is talking about _not_ touching the definition in the UAPI header. Does the rest below makes more sense now?

Quoted 7 lines

No, I'm still confused :)

We can't touch the UAPI definitions, but the fact that this api never did anything still is ok as after this patch it continues to not do anything.

I'm confused as to what you are proposing...

The UAPI definition can't be removed, but it would be nice to issue a compiler _warning_ if it's ever used.

Like eg. __attribute__ ((deprecated))

Meow!

Signature · 4 lines
⢀⣴⠾⠻⢶⣦⠀    .--[ Makefile ]
⣾⠁⢠⠒⠀⣿⡁    # beware of races
⢿⡄⠘⠷⠚⠋⠀    all: pillage burn
⠈⠳⣄⠀⠀⠀⠀    `----
#9 lore raw
Quoted 19 lines · Adam Borowski
Quoted 5 lines · Greg Kroah-Hartman
Quoted 4 lines · Jiri Slaby
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines · Jiri Slaby
Quoted 3 lines · Jann Horn
Quoted 2 lines

Delete this dead code; but leave the definition of struct termiox in the UAPI headers intact.

[was snipped]

Quoted 12 lines
Quoted 5 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

I am thinking -- can/should we mark the structure as deprecated so that userspace stops using it eventually?

Note this ^^^^^. He is talking about _not_ touching the definition in the UAPI header. Does the rest below makes more sense now?

No, I'm still confused :)

We can't touch the UAPI definitions, but the fact that this api never did anything still is ok as after this patch it continues to not do anything.

Quoted 6 lines
Quoted 1 line

I'm confused as to what you are proposing...

The UAPI definition can't be removed, but it would be nice to issue a compiler _warning_ if it's ever used.

Like eg. __attribute__ ((deprecated))

Don't add build warnings for no good reasons, that's not nice. As the feature just doesn't work, anyone who tries to use it will very quickly realize that :)

thanks,

greg k-h

#10 lore raw
In reply to Jiri Slaby
Quoted 6 lines · Jiri Slaby
Quoted 6 lines
Quoted 5 lines
Quoted 2 lines

Delete this dead code; but leave the definition of struct termiox in the UAPI headers intact.

I am thinking -- can/should we mark the structure as deprecated so that userspace stops using it eventually?

Quoted 7 lines
Quoted 1 line

If it doesn't do anything, how can userspace even use it today? :)

Well, right. I am in favor to remove it, BUT: what if someone tries that ioctl and bails out if EINVAL is returned. I mean: if they define a local var of that struct type and pass it to the ioctl, we would break the build by removing the struct completely. Even if the code didn't do anything useful, it still could be built. So is this very potential breakage OK?

Um, we do guarantee a stable ABI. We have never guaranteed that all old crappy code will continue to compile, although we avoid gratious breakage. And assuming there ever was code using termiox (which I'm not sure about to start with) it will surely have some form of feature check, and I think we are better off with that feature check not detecting the presence as that would be completely pointless.

Or in short: by keeping the uapi definition we do userspace software a disfavor.

maintainer
#11 lore raw
Quoted 22 lines · Christoph Hellwig
Quoted 6 lines · Jiri Slaby
Quoted 6 lines
Quoted 5 lines
Quoted 2 lines

Delete this dead code; but leave the definition of struct termiox in the UAPI headers intact.

I am thinking -- can/should we mark the structure as deprecated so that userspace stops using it eventually?

Quoted 7 lines
Quoted 1 line

If it doesn't do anything, how can userspace even use it today? :)

Well, right. I am in favor to remove it, BUT: what if someone tries that ioctl and bails out if EINVAL is returned. I mean: if they define a local var of that struct type and pass it to the ioctl, we would break the build by removing the struct completely. Even if the code didn't do anything useful, it still could be built. So is this very potential breakage OK?

Um, we do guarantee a stable ABI. We have never guaranteed that all old crappy code will continue to compile, although we avoid gratious breakage. And assuming there ever was code using termiox (which I'm not sure about to start with) it will surely have some form of feature check, and I think we are better off with that feature check not detecting the presence as that would be completely pointless.

Quoted 2 lines

Or in short: by keeping the uapi definition we do userspace software a disfavor.

OK, even better. I will remove it once I get to it (if noone beats me to it, of course).

thanks,

Signature · 2 lines
js
suse labs