linux-kernel - [PATCH v1 1/1] tty: tty_io: Switch to vmalloc() fallback in case of TTY_NO_WRITE

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite for Android: free password hash cracker in your pocket

[<prev] [next>] [day] [month] [year] [list]

Date:   Mon, 20 Dec 2021 15:32:50 +0200
From:   Andy Shevchenko <andriy.shevchenko@...ux.intel.com>
To:     Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
        linux-kernel@...r.kernel.org, linux-usb@...r.kernel.org
Cc:     Jiri Slaby <jirislaby@...nel.org>,
        Oliver Neukum <oneukum@...e.com>,
        Andy Shevchenko <andriy.shevchenko@...ux.intel.com>
Subject: [PATCH v1 1/1] tty: tty_io: Switch to vmalloc() fallback in case of TTY_NO_WRITE_SPLIT

When TTY_NO_WRITE_SPLIT is set and 64 KiB chunks are used, allow
vmalloc() fallback. Supply __GFP_RETRY_MAYFAIL to make kmalloc()
preferable over vmalloc() since we may want a better performance.

Note, both current users copy data to another buffer anyway, so
the type of our allocation doesn't affect their expectations.

Signed-off-by: Andy Shevchenko <andriy.shevchenko@...ux.intel.com>
---
 drivers/tty/tty_io.c        | 9 +++------
 drivers/usb/class/cdc-acm.c | 4 ----
 2 files changed, 3 insertions(+), 10 deletions(-)

diff --git a/drivers/tty/tty_io.c b/drivers/tty/tty_io.c
index 6616d4a0d41d..8fedfe88dff7 100644
--- a/drivers/tty/tty_io.c
+++ b/drivers/tty/tty_io.c
@@ -170,7 +170,7 @@ static void free_tty_struct(struct tty_struct *tty)
 {
 	tty_ldisc_deinit(tty);
 	put_device(tty->dev);
-	kfree(tty->write_buf);
+	kvfree(tty->write_buf);
 	tty->magic = 0xDEADDEAD;
 	kfree(tty);
 }
@@ -997,9 +997,6 @@ static inline ssize_t do_tty_write(
 	 * layer has problems with bigger chunks. It will
 	 * claim to be able to handle more characters than
 	 * it actually does.
-	 *
-	 * FIXME: This can probably go away now except that 64K chunks
-	 * are too likely to fail unless switched to vmalloc...
 	 */
 	chunk = 2048;
 	if (test_bit(TTY_NO_WRITE_SPLIT, &tty->flags))
@@ -1014,12 +1011,12 @@ static inline ssize_t do_tty_write(
 		if (chunk < 1024)
 			chunk = 1024;
 
-		buf_chunk = kmalloc(chunk, GFP_KERNEL);
+		buf_chunk = kvmalloc(chunk, GFP_KERNEL | __GFP_RETRY_MAYFAIL);
 		if (!buf_chunk) {
 			ret = -ENOMEM;
 			goto out;
 		}
-		kfree(tty->write_buf);
+		kvfree(tty->write_buf);
 		tty->write_cnt = chunk;
 		tty->write_buf = buf_chunk;
 	}
diff --git a/drivers/usb/class/cdc-acm.c b/drivers/usb/class/cdc-acm.c
index b3ce7338cb6b..9b9aea24d58c 100644
--- a/drivers/usb/class/cdc-acm.c
+++ b/drivers/usb/class/cdc-acm.c
@@ -685,10 +685,6 @@ static int acm_port_activate(struct tty_port *port, struct tty_struct *tty)
 	if (retval)
 		goto error_get_interface;
 
-	/*
-	 * FIXME: Why do we need this? Allocating 64K of physically contiguous
-	 * memory is really nasty...
-	 */
 	set_bit(TTY_NO_WRITE_SPLIT, &tty->flags);
 	acm->control->needs_remote_wakeup = 1;
 
-- 
2.34.1