Skip to contentSkip to conversation

um: Assorted console related fixes

PATCH In review Richard Weinberger 6 participants 19 messages Jan 1, 2022 – Jun 7, 2022 User-mode Linux (UML)

lore
Share & tools
lore

Richard Weinberger

author maintainer
#0 lore raw

cover um: Assorted console related fixes

When UML terminates (gracefully or panic) it does not always cleanup on the host side.
Especially when fd:0,fd:1 is used as console it can leave a misconfigured stdin/out on the host side.
Also stale processes such as port-helper can left behind.

This series fixes the console exit code to make sure UML undoes everything it changed on the host side.

Richard Weinberger (4):
um: Run console exit code also upon kernel panic
um: os_set_fd_block: Return old blocking mode
um: chan: Restore blocking mode upon exit
um: irq: Restore O_ASYNC upon exit

 arch/um/drivers/chan.h               |  1 +
 arch/um/drivers/chan_kern.c          | 11 +++++++----
 arch/um/drivers/chan_user.c          |  2 +-
 arch/um/drivers/mconsole_kern.c      |  2 +-
 arch/um/drivers/stdio_console.c      |  4 ++--
 arch/um/drivers/ubd_user.c           |  2 +-
 arch/um/drivers/xterm.c              |  2 +-
 arch/um/include/shared/uml_console.h | 11 +++++++++++
 arch/um/kernel/irq.c                 |  5 ++++-
 arch/um/kernel/um_arch.c             |  8 +++++---
 arch/um/os-Linux/file.c              |  5 ++++-
 11 files changed, 38 insertions(+), 15 deletions(-)
 create mode 100644 arch/um/include/shared/uml_console.h
Signature · 7 lines
2.26.2


_______________________________________________
linux-um mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-um
author maintainer
#1 lore raw

PATCH 1/4 um: Run console exit code also upon kernel panic

Make sure that UML's console exit handler run also when the kernel crashes.
That way settings and resources can get restored.

---
 arch/um/drivers/stdio_console.c      |  4 ++--
 arch/um/include/shared/uml_console.h | 11 +++++++++++
 arch/um/kernel/um_arch.c             |  8 +++++---
 3 files changed, 18 insertions(+), 5 deletions(-)
 create mode 100644 arch/um/include/shared/uml_console.h
arch/um/drivers/stdio_console.c +2 -2
index e8b762f4d8c2..06612c3b6240 100644
@@ -180,13 +180,13 @@ static int stdio_init(void)
 }
 late_initcall(stdio_init);
 
-static void console_exit(void)
+void uml_console_exit(void)
 {
 	if (!con_init_done)
 		return;
 	close_lines(vts, ARRAY_SIZE(vts));
 }
-__uml_exitcall(console_exit);
+__uml_exitcall(uml_console_exit);
 
 static int console_chan_setup(char *str)
 {
arch/um/include/shared/uml_console.h new file +11 -0
new file mode 100644
index 000000000000..aa8d02fe5e90
@@ -0,0 +1,11 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * Copyright (C) 2022 Richard Weinberger <[email protected]>
+ */
+
+#ifndef UML_CONSOLE_H
+#define UML_CONSOLE_H
+
+extern void uml_console_exit(void);
+
+#endif /* UML_CONSOLE_H */
arch/um/kernel/um_arch.c +5 -3
index abceeabe29b9..0bdae61929ef 100644
@@ -17,17 +17,18 @@
 #include <linux/kmsg_dump.h>
 #include <linux/suspend.h>
 
-#include <asm/processor.h>
+#include <arch.h>
+#include <as-layout.h>
 #include <asm/cpufeature.h>
+#include <asm/processor.h>
 #include <asm/sections.h>
 #include <asm/setup.h>
-#include <as-layout.h>
-#include <arch.h>
 #include <init.h>
 #include <kern.h>
 #include <kern_util.h>
 #include <mem_user.h>
 #include <os.h>
+#include <uml_console.h>
 
 #include "um_arch.h"
 
@@ -245,6 +246,7 @@ static int panic_exit(struct notifier_block *self, unsigned long unused1,
 	bust_spinlocks(1);
 	bust_spinlocks(0);
 	uml_exitcode = 1;
+	uml_console_exit();
 	os_dump_core();
 	return 0;
 }
Signature · 1 line
2.26.2
#2 lore raw

Re: [PATCH 1/4] um: Run console exit code also upon kernel panic

I love your patch! Perhaps something to improve:

Hi Richard,

I love your patch! Perhaps something to improve:

[auto build test WARNING on rw-uml/linux-next]
[also build test WARNING on linux/master linus/master v5.16-rc8 next-20211224] [If your patch is applied to the wrong git tree, kindly drop us a note. And when submitting patch, we suggest to use '--base' as documented in
https://git-scm.com/docs/git-format-patch]

url:    https://github.com/0day-ci/linux/commits/Richard-Weinberger/um-Assorted-console-related-fixes/20220102-055954
base:   https://git.kernel.org/pub/scm/linux/kernel/git/rw/uml.git linux-next
config: um-i386_defconfig (https://download.01.org/0day-ci/archive/20220104/202201041330.hsCRX4ov-lkp(a)intel.com/config)
compiler: gcc-9 (Debian 9.3.0-22) 9.3.0

reproduce (this is a W=1 build):
# https://github.com/0day-ci/linux/commit/8ed1e739d9071fbe06c0e6ff91eeb065afd18953
git remote add linux-review https://github.com/0day-ci/linux
git fetch --no-tags linux-review Richard-Weinberger/um-Assorted-console-related-fixes/20220102-055954 git checkout 8ed1e739d9071fbe06c0e6ff91eeb065afd18953
# save the config file to linux build tree
mkdir build_dir
make W=1 O=build_dir ARCH=um SUBARCH=i386 SHELL=/bin/bash

If you fix the issue, kindly add following tag as appropriate

All warnings (new ones prefixed by >>):

>> arch/um/drivers/stdio_console.c:183:6: warning: no previous prototype for 'uml_console_exit' [-Wmissing-prototypes]

183 | void uml_console_exit(void)
| ^~~~~~~~~~~~~~~~

vim +/uml_console_exit +183 arch/um/drivers/stdio_console.c

182

 > 183	void uml_console_exit(void)

184 {
185 if (!con_init_done)
186 return;
187 close_lines(vts, ARRAY_SIZE(vts));
188 }
189 __uml_exitcall(uml_console_exit);
190

---
0-DAY CI Kernel Test Service, Intel Corporation
https://lists.01.org/hyperkitty/list/kbuild-all(a)lists.01.org

maintainer
#3 lore raw

Re: [PATCH 1/4] um: Run console exit code also upon kernel panic

In reply to Richard Weinberger
Quoted 8 lines · Richard Weinberger
arch/um/include/shared/uml_console.h new file +0 -0
new file mode 100644
index 000000000000..aa8d02fe5e90

arch/um/include/shared/uml_console.h line 9

Quoted 9 lines
arch/um/include/shared/uml_console.h new file +9 -0
@@ -0,0 +1,11 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * Copyright (C) 2022 Richard Weinberger <[email protected]>
+ */
+
+#ifndef UML_CONSOLE_H
+#define UML_CONSOLE_H
+
+extern void uml_console_exit(void);

Not sure it's worth adding a whole new file just for this, but also doesn't really matter :)

johannes

_______________________________________________
linux-um mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-um

author maintainer
#4 lore raw

PATCH 2/4 um: os_set_fd_block: Return old blocking mode

In reply to Richard Weinberger

This will be needed when we restore the old mode upon exit. The function now returns < 0 in case of an error, 1 if O_NONBLOCK was set, 0 otherwise.

---
 arch/um/drivers/chan_kern.c     | 8 ++++----
 arch/um/drivers/chan_user.c     | 2 +-
 arch/um/drivers/mconsole_kern.c | 2 +-
 arch/um/drivers/ubd_user.c      | 2 +-
 arch/um/drivers/xterm.c         | 2 +-
 arch/um/os-Linux/file.c         | 5 ++++-
 6 files changed, 12 insertions(+), 9 deletions(-)
arch/um/drivers/chan_kern.c +4 -4
index 62997055c454..86be115b5890 100644
@@ -83,7 +83,7 @@ static const struct chan_ops not_configged_ops = {
 
 static int open_one_chan(struct chan *chan)
 {
-	int fd, err;
+	int fd, ret;
 
 	if (chan->opened)
 		return 0;
@@ -95,10 +95,10 @@ static int open_one_chan(struct chan *chan)
 	if (fd < 0)
 		return fd;
 
-	err = os_set_fd_block(fd, 0);
-	if (err) {
+	ret = os_set_fd_block(fd, 0);
+	if (ret < 0) {
 		(*chan->ops->close)(fd, chan->data);
-		return err;
+		return ret;
 	}
 
 	chan->fd = fd;
arch/um/drivers/chan_user.c +1 -1
index 6040817c036f..b449656cd5f8 100644
@@ -257,7 +257,7 @@ static int winch_tramp(int fd, struct tty_port *port, int *fd_out,
 	}
 
 	err = os_set_fd_block(*fd_out, 0);
-	if (err) {
+	if (err < 0) {
 		printk(UM_KERN_ERR "winch_tramp: failed to set thread_fd "
 		       "non-blocking.\n");
 		goto out_close;
arch/um/drivers/mconsole_kern.c +1 -1
index 6ead1e240457..8000914f9cd4 100644
@@ -732,7 +732,7 @@ static int __init mconsole_init(void)
 		printk(KERN_ERR "Failed to initialize management console\n");
 		return 1;
 	}
-	if (os_set_fd_block(sock, 0))
+	if (os_set_fd_block(sock, 0) < 0)
 		goto out;
 
 	register_reboot_notifier(&reboot_notifier);
arch/um/drivers/ubd_user.c +1 -1
index a1afe414ce48..a930a254826b 100644
@@ -42,7 +42,7 @@ int start_io_thread(unsigned long sp, int *fd_out)
 
 	err = os_set_fd_block(*fd_out, 0);
 	err = os_set_fd_block(kernel_fd, 0);
-	if (err) {
+	if (err < 0) {
 		printk("start_io_thread - failed to set nonblocking I/O.\n");
 		goto out_close;
 	}
arch/um/drivers/xterm.c +1 -1
index 87ca4a47cd66..01be81c25266 100644
@@ -161,7 +161,7 @@ static int xterm_open(int input, int output, int primary, void *d,
 	}
 
 	err = os_set_fd_block(new, 0);
-	if (err) {
+	if (err < 0) {
 		printk(UM_KERN_ERR "xterm_open : failed to set xterm "
 		       "descriptor non-blocking, err = %d\n", -err);
 		goto out_close2;
arch/um/os-Linux/file.c +4 -1
index e4421dbc4c36..af2b3cba2c93 100644
@@ -445,11 +445,14 @@ int os_clear_fd_async(int fd)
 int os_set_fd_block(int fd, int blocking)
 {
 	int flags;
+	int oldmode;
 
 	flags = fcntl(fd, F_GETFL);
 	if (flags < 0)
 		return -errno;
 
+	oldmode = !!(flags & O_NONBLOCK);
+
 	if (blocking)
 		flags &= ~O_NONBLOCK;
 	else
@@ -458,7 +461,7 @@ int os_set_fd_block(int fd, int blocking)
 	if (fcntl(fd, F_SETFL, flags) < 0)
 		return -errno;
 
-	return 0;
+	return oldmode;
 }
 
 int os_accept_connection(int fd)
Signature · 7 lines
2.26.2


_______________________________________________
linux-um mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-um
maintainer
#5 lore raw

Re: [PATCH 2/4] um: os_set_fd_block: Return old blocking mode

Quoted 6 lines · Richard Weinberger

This will be needed when we restore the old mode upon exit. The function now returns < 0 in case of an error, 1 if O_NONBLOCK was set, 0 otherwise.

+++ b/arch/um/drivers/chan_user.c

arch/um/drivers/chan_user.c static int winch_tramp(int fd, struct tty_port *port, int *fd_out, · line 263

Quoted 8 lines
arch/um/drivers/chan_user.c +1 -1
@@ -257,7 +257,7 @@ static int winch_tramp(int fd, struct tty_port *port, int *fd_out,
 	}
 
 	err = os_set_fd_block(*fd_out, 0);
-	if (err) {
+	if (err < 0) {
 		printk(UM_KERN_ERR "winch_tramp: failed to set thread_fd "
 		       "non-blocking.\n");
 		goto out_close;

This does 'return err' afterwards, which then changes the logic from 0 to 1, which seems wrong - especially given that it gets passed further elsewhere as a pid, which seems very questionable in the first place, but at least 1 would be valid pid unlike 0?

IOW, I think you need

- return err;
+ return 0;

a couple of lines later.

johannes

_______________________________________________
linux-um mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-um

author maintainer
#6 lore raw

PATCH 3/4 um: chan: Restore blocking mode upon exit

In reply to Richard Weinberger

Make sure that O_NONBLOCK is cleared upon exit if UML sets it. Otherwise stdin/out behaves strange after UML terminates.

---
 arch/um/drivers/chan.h      | 1 +
 arch/um/drivers/chan_kern.c | 3 +++
 2 files changed, 4 insertions(+)
arch/um/drivers/chan.h +1 -0
index c37cc4f26f91..6b8c462f146e 100644
@@ -22,6 +22,7 @@ struct chan {
 	unsigned int output:1;
 	unsigned int opened:1;
 	unsigned int enabled:1;
+	unsigned int fd_blocking:1;
 	int fd;
 	const struct chan_ops *ops;
 	void *data;
arch/um/drivers/chan_kern.c +3 -0
index 86be115b5890..6d512589fa50 100644
@@ -102,6 +102,7 @@ static int open_one_chan(struct chan *chan)
 	}
 
 	chan->fd = fd;
+	chan->fd_blocking = ret;
 
 	chan->opened = 1;
 	return 0;
@@ -223,6 +224,8 @@ static void close_one_chan(struct chan *chan, int delay_free_irq)
 	if (chan->ops->close != NULL)
 		(*chan->ops->close)(chan->fd, chan->data);
 
+	os_set_fd_block(chan->fd, chan->fd_blocking);
+
 	chan->opened = 0;
 	chan->fd = -1;
 }
Signature · 1 line
2.26.2
author maintainer
#7 lore raw

PATCH 4/4 um: irq: Restore O_ASYNC upon exit

In reply to Richard Weinberger

Just like with O_NONBLOCK, make sure UML clears O_ASYNC upon exit when it set it.

---
 arch/um/kernel/irq.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)
arch/um/kernel/irq.c +4 -1
index a8873d9bc28b..e020bdccfa23 100644
@@ -342,6 +342,7 @@ static void free_irq_by_irq_and_dev(unsigned int irq, void *dev)
 				continue;
 
 			os_del_epoll_fd(entry->fd);
+			os_clear_fd_async(entry->fd);
 			reg->events = 0;
 			update_or_free_irq_entry(entry);
 			goto out;
@@ -396,8 +397,10 @@ int deactivate_all_fds(void)
 	os_set_ioignore();
 
 	/* we can no longer call kfree() here so just deactivate */
-	list_for_each_entry(entry, &active_fds, list)
+	list_for_each_entry(entry, &active_fds, list) {
 		os_del_epoll_fd(entry->fd);
+		os_clear_fd_async(entry->fd);
+	}
 	os_close_epoll_fd();
 	return 0;
 }
Signature · 7 lines
2.26.2


_______________________________________________
linux-um mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-um
maintainer
#8 lore raw

Re: [PATCH 4/4] um: irq: Restore O_ASYNC upon exit

Quoted 3 lines · Richard Weinberger

Just like with O_NONBLOCK, make sure UML clears O_ASYNC upon exit when it set it.

"when it set it" isn't really quite true - you always clear it, should it be similar to the O_NONBLOCK version to see the old mode?

johannes

_______________________________________________
linux-um mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-um

#9 lore raw

cover um: Assorted console related fixes

In reply to Richard Weinberger
Quoted 8 lines · Richard Weinberger

When UML terminates (gracefully or panic) it does not always cleanup on the host side.
Especially when fd:0,fd:1 is used as console it can leave a misconfigured stdin/out on the host side.
Also stale processes such as port-helper can left behind.

This series fixes the console exit code to make sure UML undoes everything it changed on the host side.

I gave this a run hoping it would fix the issues I'm seeing with the console getting messed up when UML exits via halt in the guest. Echo is turned off and looks like carriage return doesn't work either. I'm using "reset" to get the terminal back to its previous state. Unfortunately this patch series did not fix this issue.

I'm not saying that these aren't valid fixes. But it does appear that these changes don't make sure that "everything" changed on the host side is undone.

Glenn

_______________________________________________
linux-um mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-um

author maintainer
#10 lore raw

----- Ursprüngliche Mail -----

Quoted 11 lines
Quoted 2 lines

This series fixes the console exit code to make sure UML undoes everything it changed on the host side.

I gave this a run hoping it would fix the issues I'm seeing with the console getting messed up when UML exits via halt in the guest. Echo is turned off and looks like carriage return doesn't work either. I'm using "reset" to get the terminal back to its previous state. Unfortunately this patch series did not fix this issue.

I'm not saying that these aren't valid fixes. But it does appear that these changes don't make sure that "everything" changed on the host side is undone.

So when you compare "stty -a" before and after the damange you still see differences? Please also compare the flags of /proc/$$/fdinfo/{0,1,2}

Thanks,
//richard

_______________________________________________
linux-um mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-um

#11 lore raw
Quoted 10 lines · Richard Weinberger

----- Ursprüngliche Mail -----

Quoted 9 lines
Quoted 2 lines

This series fixes the console exit code to make sure UML undoes everything it changed on the host side.

I gave this a run hoping it would fix the issues I'm seeing with the console getting messed up when UML exits via halt in the guest. Echo is turned off and looks like carriage return doesn't work either. I'm using "reset" to get the terminal back to its previous state. Unfortunately this patch series did not fix this issue.

Quoted 5 lines
Quoted 2 lines

I'm not saying that these aren't valid fixes. But it does appear that these changes don't make sure that "everything" changed on the host side is undone.

So when you compare "stty -a" before and after the damange you still see differences? Please also compare the flags of /proc/$$/fdinfo/{0,1,2}

Uh, where are the patches?

Quoted 1 line

http://lists.infradead.org/mailman/listinfo/linux-um

not in that archive ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

Signature · 6 lines
~Randy

_______________________________________________
linux-um mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-um
author maintainer
#12 lore raw

Randy,

----- Ursprüngliche Mail -----

Quoted 6 lines

Von: "Randy Dunlap" <[email protected]>
Uh, where are the patches?

Quoted 1 line

http://lists.infradead.org/mailman/listinfo/linux-um

not in that archive ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

https://lore.kernel.org/lkml/[email protected]/T/

Thanks,
//richard

_______________________________________________
linux-um mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-um

#13 lore raw
Quoted 11 lines · Richard Weinberger

Randy,

----- Ursprüngliche Mail -----

Quoted 6 lines

Von: "Randy Dunlap" <[email protected]>
Uh, where are the patches?

Quoted 1 line

http://lists.infradead.org/mailman/listinfo/linux-um

not in that archive ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

https://lore.kernel.org/lkml/[email protected]/T/

Thanks.

Signature · 6 lines
~Randy

_______________________________________________
linux-um mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-um
#14 lore raw
In reply to Richard Weinberger
Quoted 10 lines · On Wed, 9 Mar 2022 09:35:02 +0100 (CET) Richard Weinberger <[email protected]>…

----- Ursprüngliche Mail -----

Quoted 9 lines
Quoted 2 lines

This series fixes the console exit code to make sure UML undoes everything it changed on the host side.

I gave this a run hoping it would fix the issues I'm seeing with the console getting messed up when UML exits via halt in the guest. Echo is turned off and looks like carriage return doesn't work either. I'm using "reset" to get the terminal back to its previous state. Unfortunately this patch series did not fix this issue.

Quoted 5 lines
Quoted 2 lines

I'm not saying that these aren't valid fixes. But it does appear that these changes don't make sure that "everything" changed on the host side is undone.

So when you compare "stty -a" before and after the damange you still see differences? Please also compare the flags of /proc/$$/fdinfo/{0,1,2}

Here's a diff of the before and after:
-- /tmp/uml-console.before.out 2022-03-10 18:12:54.049171650 +0000 +++ /tmp/uml-console.after.out 2022-03-10 18:13:20.711630281 +0000 @@ -3,20 +3,20 @@
eol2 = <undef>; swtch = <undef>; start = ^Q; stop = ^S; susp = ^Z; rprnt = ^R; werase = ^W; lnext = ^V; discard = ^O; min = 1; time = 0;
-parenb -parodd -cmspar cs8 hupcl -cstopb cread -clocal -crtscts
--ignbrk brkint ignpar -parmrk -inpck -istrip -inlcr -igncr icrnl ixon -ixoff +-ignbrk -brkint ignpar -parmrk -inpck -istrip -inlcr -igncr -icrnl -ixon -ixoff
-iuclc -ixany imaxbel -iutf8
-opost -olcuc -ocrnl onlcr -onocr -onlret -ofill -ofdel nl0 cr0 tab0 bs0 vt0 ff0 -isig icanon -iexten echo echoe echok -echonl -noflsh -xcase -tostop -echoprt

+-opost -olcuc -ocrnl onlcr -onocr -onlret -ofill -ofdel nl0 cr0 tab0 bs0 vt0 ff0
+-isig -icanon -iexten -echo echoe echok -echonl -noflsh -xcase -tostop -echoprt
 echoctl echoke -flusho -extproc
 /proc/3383323/fdinfo/0:1:pos:  0
-/proc/3383323/fdinfo/0:2:flags:        0120002
+/proc/3383323/fdinfo/0:2:flags:        0124002
 /proc/3383323/fdinfo/0:3:mnt_id:       6104
 /proc/3383323/fdinfo/0:4:ino:  24
 /proc/3383323/fdinfo/1:1:pos:  0
-/proc/3383323/fdinfo/1:2:flags:        0120002
+/proc/3383323/fdinfo/1:2:flags:        0124002
 /proc/3383323/fdinfo/1:3:mnt_id:       6104
 /proc/3383323/fdinfo/1:4:ino:  24
 /proc/3383323/fdinfo/2:1:pos:  0
-/proc/3383323/fdinfo/2:2:flags:        0120002
+/proc/3383323/fdinfo/2:2:flags:        0124002
 /proc/3383323/fdinfo/2:3:mnt_id:       6104

/proc/3383323/fdinfo/2:4:ino: 24

It looks like these were not reset: brkin icrnl ixon opost isig icanon echo

Originally I was applying this patch series to 5.16.12 and seeing this issue.
But I this output is from applying the patch series to 5.17-rc7, if that matters. Thanks for looking t this.

Glenn

Quoted 3 lines

Thanks,
//richard

_______________________________________________
linux-um mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-um

#15 lore raw

Glenn,

Quoted 51 lines · Glenn Washburn
Quoted 10 lines · On Wed, 9 Mar 2022 09:35:02 +0100 (CET) Richard Weinberger <[email protected]>…

----- Ursprüngliche Mail -----

Quoted 9 lines
Quoted 2 lines

This series fixes the console exit code to make sure UML undoes everything it changed on the host side.

I gave this a run hoping it would fix the issues I'm seeing with the console getting messed up when UML exits via halt in the guest. Echo is turned off and looks like carriage return doesn't work either. I'm using "reset" to get the terminal back to its previous state. Unfortunately this patch series did not fix this issue.

Quoted 5 lines
Quoted 2 lines

I'm not saying that these aren't valid fixes. But it does appear that these changes don't make sure that "everything" changed on the host side is undone.

So when you compare "stty -a" before and after the damange you still see differences? Please also compare the flags of /proc/$$/fdinfo/{0,1,2}

Here's a diff of the before and after:
-- /tmp/uml-console.before.out 2022-03-10 18:12:54.049171650 +0000 +++ /tmp/uml-console.after.out 2022-03-10 18:13:20.711630281 +0000

(quoted diff) +6 -6
@@ -3,20 +3,20 @@
 eol2 = <undef>; swtch = <undef>; start = ^Q; stop = ^S; susp = ^Z; rprnt = ^R;
 werase = ^W; lnext = ^V; discard = ^O; min = 1; time = 0;
 -parenb -parodd -cmspar cs8 hupcl -cstopb cread -clocal -crtscts
--ignbrk brkint ignpar -parmrk -inpck -istrip -inlcr -igncr icrnl ixon -ixoff
+-ignbrk -brkint ignpar -parmrk -inpck -istrip -inlcr -igncr -icrnl -ixon -ixoff
 -iuclc -ixany imaxbel -iutf8
-opost -olcuc -ocrnl onlcr -onocr -onlret -ofill -ofdel nl0 cr0 tab0 bs0 vt0 ff0
-isig icanon -iexten echo echoe echok -echonl -noflsh -xcase -tostop -echoprt
+-opost -olcuc -ocrnl onlcr -onocr -onlret -ofill -ofdel nl0 cr0 tab0 bs0 vt0 ff0
+-isig -icanon -iexten -echo echoe echok -echonl -noflsh -xcase -tostop -echoprt
 echoctl echoke -flusho -extproc
 /proc/3383323/fdinfo/0:1:pos:  0
-/proc/3383323/fdinfo/0:2:flags:        0120002
+/proc/3383323/fdinfo/0:2:flags:        0124002
 /proc/3383323/fdinfo/0:3:mnt_id:       6104
 /proc/3383323/fdinfo/0:4:ino:  24
 /proc/3383323/fdinfo/1:1:pos:  0
-/proc/3383323/fdinfo/1:2:flags:        0120002
+/proc/3383323/fdinfo/1:2:flags:        0124002
 /proc/3383323/fdinfo/1:3:mnt_id:       6104
 /proc/3383323/fdinfo/1:4:ino:  24
 /proc/3383323/fdinfo/2:1:pos:  0
-/proc/3383323/fdinfo/2:2:flags:        0120002
+/proc/3383323/fdinfo/2:2:flags:        0124002
 /proc/3383323/fdinfo/2:3:mnt_id:       6104
 /proc/3383323/fdinfo/2:4:ino:  24
Quoted 4 lines

It looks like these were not reset: brkin icrnl ixon opost isig icanon echo

Originally I was applying this patch series to 5.16.12 and seeing this issue.
But I this output is from applying the patch series to 5.17-rc7, if that matters. Thanks for looking t this.

Hmm. Looks like the cleanup function is never called. Can you verify whether uml_console_exit() runs at exit time?

Signature · 7 lines
Thanks,
//richard

_______________________________________________
linux-um mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-um
#16 lore raw
Quoted 55 lines · On Mon, 4 Apr 2022 23:41:29 +0200 Richard Weinberger…

Glenn,

Quoted 51 lines · Glenn Washburn
Quoted 10 lines · On Wed, 9 Mar 2022 09:35:02 +0100 (CET) Richard Weinberger <[email protected]>…

----- Ursprüngliche Mail -----

Quoted 9 lines
Quoted 2 lines

This series fixes the console exit code to make sure UML undoes everything it changed on the host side.

I gave this a run hoping it would fix the issues I'm seeing with the console getting messed up when UML exits via halt in the guest. Echo is turned off and looks like carriage return doesn't work either. I'm using "reset" to get the terminal back to its previous state. Unfortunately this patch series did not fix this issue.

Quoted 5 lines
Quoted 2 lines

I'm not saying that these aren't valid fixes. But it does appear that these changes don't make sure that "everything" changed on the host side is undone.

So when you compare "stty -a" before and after the damange you still see differences? Please also compare the flags of /proc/$$/fdinfo/{0,1,2}

Here's a diff of the before and after:
-- /tmp/uml-console.before.out 2022-03-10 18:12:54.049171650 +0000 +++ /tmp/uml-console.after.out 2022-03-10 18:13:20.711630281 +0000

(quoted diff) +6 -6
@@ -3,20 +3,20 @@
 eol2 = <undef>; swtch = <undef>; start = ^Q; stop = ^S; susp = ^Z; rprnt = ^R;
 werase = ^W; lnext = ^V; discard = ^O; min = 1; time = 0;
 -parenb -parodd -cmspar cs8 hupcl -cstopb cread -clocal -crtscts
--ignbrk brkint ignpar -parmrk -inpck -istrip -inlcr -igncr icrnl ixon -ixoff
+-ignbrk -brkint ignpar -parmrk -inpck -istrip -inlcr -igncr -icrnl -ixon -ixoff
 -iuclc -ixany imaxbel -iutf8
-opost -olcuc -ocrnl onlcr -onocr -onlret -ofill -ofdel nl0 cr0 tab0 bs0 vt0 ff0
-isig icanon -iexten echo echoe echok -echonl -noflsh -xcase -tostop -echoprt
+-opost -olcuc -ocrnl onlcr -onocr -onlret -ofill -ofdel nl0 cr0 tab0 bs0 vt0 ff0
+-isig -icanon -iexten -echo echoe echok -echonl -noflsh -xcase -tostop -echoprt
 echoctl echoke -flusho -extproc
 /proc/3383323/fdinfo/0:1:pos:  0
-/proc/3383323/fdinfo/0:2:flags:        0120002
+/proc/3383323/fdinfo/0:2:flags:        0124002
 /proc/3383323/fdinfo/0:3:mnt_id:       6104
 /proc/3383323/fdinfo/0:4:ino:  24
 /proc/3383323/fdinfo/1:1:pos:  0
-/proc/3383323/fdinfo/1:2:flags:        0120002
+/proc/3383323/fdinfo/1:2:flags:        0124002
 /proc/3383323/fdinfo/1:3:mnt_id:       6104
 /proc/3383323/fdinfo/1:4:ino:  24
 /proc/3383323/fdinfo/2:1:pos:  0
-/proc/3383323/fdinfo/2:2:flags:        0120002
+/proc/3383323/fdinfo/2:2:flags:        0124002
 /proc/3383323/fdinfo/2:3:mnt_id:       6104
 /proc/3383323/fdinfo/2:4:ino:  24
Quoted 7 lines
Quoted 4 lines

It looks like these were not reset: brkin icrnl ixon opost isig icanon echo

Originally I was applying this patch series to 5.16.12 and seeing this issue.
But I this output is from applying the patch series to 5.17-rc7, if that matters. Thanks for looking t this.

Hmm. Looks like the cleanup function is never called. Can you verify whether uml_console_exit() runs at exit time?

Sorry about the delay. I don't think that uml_console_exit() is getting run. I'm testing this by running a UML instance which has been compiled with -ggdb3. Then attach to the running process via GDB. Set a breakpoint on uml_console_exit(), then do a shutdown/halt (syscall reboot(0x4321fedc)). GDB says that the process was terminated with SIGKILL, which I'm guessing is why uml_console_exit() is not run.

Glenn

_______________________________________________
linux-um mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-um

#18 lore raw
Quoted 6 lines · On Tue, 07 Jun 2022 10:22:43 +0200 Johannes Berg <[email protected]>…
Quoted 3 lines · Glenn Washburn

GDB says that the process was terminated with
SIGKILL, which I'm guessing is why uml_console_exit() is not run.

https://patchwork.ozlabs.org/project/linux-um/patch/20220520194536.1acbb414a2d1.I1d71bab01d45f7f8e446df6e98547a7989338208@changeid/

Great! I've tested that this solves the SIGKILL issue and that stty reports the same values before and after running a UML instance.

However, the flags re not the same (before 32770 and after 34818, in decimal). It seems that the O_NONBLOCK flag is not being restored to its original value (off).

So now the terminal is not messed up, but the flags are still an issue for me because having stdin/stdout in non-blocking mode breaks some of my scripts using uml.

Glenn

_______________________________________________
linux-um mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-um