linux-kernel - [SERIAL]fix serial_cs I/O windows for Argosy RS-COM 2P

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: Windows password security audit tool. GUI, reports in PDF.

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

Message-ID: <4ED7F40A.4040001@o2.pl>
Date:	Thu, 01 Dec 2011 22:39:22 +0100
From:	Maciej Szmigiero <mhej@...pl>
To:	Alan Cox <alan@...ux.intel.com>, linux-serial@...r.kernel.org,
	linux-kernel@...r.kernel.org
Subject: [SERIAL]fix serial_cs I/O windows for Argosy RS-COM 2P

Current serial_cs driver has a problem when trying to detect whether
a card has multiple ports: serial_config() calls pcmcia_loop_config()
which iterates over card CIS configurations by calling
serial_check_for_multi() for each of them.

This function wants to check (and select) a configuration
that has either one long I/O window spanning multiple ports or two 8-port
windows for two serial ports.

Problem is, that every pcmcia_loop_config() iteration only updates
the windows (via pcmcia_do_loop_config() in resource[0] and resource[1])
when CONF_AUTO_SET_IO flag is set on the device, which is set only later
in the code.

Fix it by setting this flag earlier.

In addition to this, when multi-port card is detected
and it does not have an one, long I/O window
multi_config_check_notpicky() tries to locate two I/O windows and assumes
they are continuous without checking.

On an Argosy RS-COM 2P this selects first configuration, which
unfortunately has two non-continuous I/O windows.
The net effect is that the second serial port on the card does not work.

Fix it by checking whether the windows are really continuous.

Signed-off-by: Maciej Szmigiero <mhej@...pl>

diff --git a/drivers/tty/serial/serial_cs.c b/drivers/tty/serial/serial_cs.c
index eef736f..8609060 100644
--- a/drivers/tty/serial/serial_cs.c
+++ b/drivers/tty/serial/serial_cs.c
@@ -317,7 +317,7 @@ static int serial_probe(struct pcmcia_device *link)
 	info->p_dev = link;
 	link->priv = info;

-	link->config_flags |= CONF_ENABLE_IRQ;
+	link->config_flags |= CONF_ENABLE_IRQ | CONF_AUTO_SET_IO;
 	if (do_sound)
 		link->config_flags |= CONF_ENABLE_SPKR;

@@ -445,7 +445,7 @@ static int simple_config(struct pcmcia_device *link)

 	/* First pass: look for a config entry that looks normal.
 	 * Two tries: without IO aliases, then with aliases */
-	link->config_flags |= CONF_AUTO_SET_VPP | CONF_AUTO_SET_IO;
+	link->config_flags |= CONF_AUTO_SET_VPP;
 	for (try = 0; try < 4; try++)
 		if (!pcmcia_loop_config(link, simple_config_check, &try))
 			goto found_port;
@@ -501,7 +501,8 @@ static int multi_config_check_notpicky(struct pcmcia_device *p_dev,
 {
 	int *base2 = priv_data;

-	if (!p_dev->resource[0]->end || !p_dev->resource[1]->end)
+	if (!p_dev->resource[0]->end || !p_dev->resource[1]->end ||
+		p_dev->resource[0]->start + 8 != p_dev->resource[1]->start)
 		return -ENODEV;

 	p_dev->resource[0]->end = p_dev->resource[1]->end = 8;
@@ -520,7 +521,6 @@ static int multi_config(struct pcmcia_device *link)
 	struct serial_info *info = link->priv;
 	int i, base2 = 0;

-	link->config_flags |= CONF_AUTO_SET_IO;
 	/* First, look for a generic full-sized window */
 	if (!pcmcia_loop_config(link, multi_config_check, &info->multi))
 		base2 = link->resource[0]->start + 8;
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/