LinuxLists.cc - 2.6.15-rc1-mm2 unsusable on DELL Inspiron 8200, 2.6.15-rc1 works fine

[permalink] [raw]

Subject: Re: 2.6.15-rc1-mm2 unsusable on DELL Inspiron 8200, 2.6.15-rc1 works fine

* Hugh Dickins <[email protected]> [2005-11-18 19:28:09 +0000]:

> On Fri, 18 Nov 2005, Marc Koschewski wrote:
> >
> > I didn't manage to read all the archive yet
>
> That should only keep you busy for a year or two!
>
> > and I'm new to the list
> > anyways. I wanted to report that 2.6.15-rc1-mm2 is unusable in an X
> > enviroment due to the mouse pointer hanging, jumping or not even working
> > at all. The behaviour occurs randomly.
>
> Many thanks for the report. It looks from the messages at the end as
> if this is a manifestation of my PageReserved "Bad page states" that
> 2.6.15-rc1-mm2 was particularly put out to check for.
>
> Unfortunately your logging system is cutting out the most important info
> (perhaps because that's marked KERN_EMERG and the stack trace is not):
> I get lost in these logging setups myself so can't tell you what to do
> to correct it: but maybe running "dmesg" will give the full output.
>
> However, I've got four other reporters (though you're the first to
> mention the mouse), and hope to be able to work it out from their info,
> so don't worry if you can't track down the full Bad page state messages.
>
> Thanks,
> Hugh

Hi Hugh,

I'm glad the report was at least a little useful. To be 'more precise':
I ran dmesg at the time all the stuff appeared in any console I had open
at that time. In between the enormous amount of lines of data the kernel
told me something about 'reboot needed' as well as the kernel tries to
cleanup something. Maybe that helps you, too. I just grep'ing the sources
to locate the printk() spiting the messages...

I located the message at mm/page_alloc.c:149 in bad_page(), just as you
told. ;)

HTH,
Marc

>
> > Nov 18 13:05:12 stiffy kernel: [bad_page+134/318] bad_page+0x86/0x13e
> > Nov 18 13:05:12 stiffy kernel: [free_hot_cold_page+66/285] free_hot_cold_page+0x42/0x11d
> > Nov 18 13:05:13 stiffy kernel: [do_gettimeofday+20/188] do_gettimeofday+0x14/0xbc
> > Nov 18 13:05:13 stiffy kernel: [zap_pte_range+305/719] zap_pte_range+0x131/0x2cf
> > Nov 18 13:05:13 stiffy kernel: [unmap_page_range+236/273] unmap_page_range+0xec/0x111
> > Nov 18 13:05:13 stiffy kernel: [unmap_vmas+193/476] unmap_vmas+0xc1/0x1dc
> > Nov 18 13:05:13 stiffy kernel: [unmap_region+150/289] unmap_region+0x96/0x121
> > Nov 18 13:05:13 stiffy kernel: [do_munmap+235/300] do_munmap+0xeb/0x12c
> > Nov 18 13:05:13 stiffy kernel: [sys_munmap+62/88] sys_munmap+0x3e/0x58
> > Nov 18 13:05:13 stiffy kernel: [syscall_call+7/11] syscall_call+0x7/0xb
> > Nov 18 13:05:14 stiffy kernel: [bad_page+134/318] bad_page+0x86/0x13e
> > Nov 18 13:05:14 stiffy kernel: [free_hot_cold_page+66/285] free_hot_cold_page+0x42/0x11d
> > ....
>

2005-11-19 03:07:26

[permalink] [raw]

Subject: Re: 2.6.15-rc1-mm2 unsusable on DELL Inspiron 8200, 2.6.15-rc1 works fine

On Friday 18 November 2005 13:29, Marc Koschewski wrote:
> Nov 18 12:58:37 stiffy kernel: psmouse.c: Wheel Mouse at isa0060/serio1/input0 lost synchronization, throwing 1 bytes away.
>
> SOME STUFF MISSING? HUH?
>
> Nov 18 13:03:14 stiffy kernel: psmouse.c: resync failed, issuing reconnect request
>

Hm, this worries me a bit... Could you please try appying the patch
below to plain 2.6.15-rc1 and see if mouse starts misbehaving again?

--
Dmitry

Input: attempt to re-synchronize mouse every 5 seconds

This should help driver to deal vith KVMs that reset mice when
switching between boxes.

Signed-off-by: Dmitry Torokhov <[email protected]>
---

drivers/input/mouse/alps.c | 37 ++++
drivers/input/mouse/logips2pp.c | 2
drivers/input/mouse/psmouse-base.c | 286 ++++++++++++++++++++++++++++---------
drivers/input/mouse/psmouse.h | 7
drivers/input/mouse/synaptics.c | 2
5 files changed, 269 insertions(+), 65 deletions(-)

Index: work/drivers/input/mouse/psmouse.h
===================================================================
--- work.orig/drivers/input/mouse/psmouse.h
+++ work/drivers/input/mouse/psmouse.h
@@ -7,7 +7,7 @@
#define PSMOUSE_CMD_GETINFO 0x03e9
#define PSMOUSE_CMD_SETSTREAM 0x00ea
#define PSMOUSE_CMD_SETPOLL 0x00f0
-#define PSMOUSE_CMD_POLL 0x03eb
+#define PSMOUSE_CMD_POLL 0x00eb /* caller sets number of bytes to receive */
#define PSMOUSE_CMD_GETID 0x02f2
#define PSMOUSE_CMD_SETRATE 0x10f3
#define PSMOUSE_CMD_ENABLE 0x00f4
@@ -23,6 +23,7 @@
enum psmouse_state {
PSMOUSE_IGNORE,
PSMOUSE_INITIALIZING,
+ PSMOUSE_RESYNCING,
PSMOUSE_CMD_MODE,
PSMOUSE_ACTIVATED,
};
@@ -38,9 +39,11 @@ struct psmouse {
void *private;
struct input_dev *dev;
struct ps2dev ps2dev;
+ struct work_struct resync_work;
char *vendor;
char *name;
unsigned char packet[8];
+ unsigned char badbyte;
unsigned char pktcnt;
unsigned char pktsize;
unsigned char type;
@@ -54,6 +57,7 @@ struct psmouse {
unsigned int rate;
unsigned int resolution;
unsigned int resetafter;
+ unsigned int resync_time;
unsigned int smartscroll; /* Logitech only */

psmouse_ret_t (*protocol_handler)(struct psmouse *psmouse, struct pt_regs *regs);
@@ -62,6 +66,7 @@ struct psmouse {

int (*reconnect)(struct psmouse *psmouse);
void (*disconnect)(struct psmouse *psmouse);
+ int (*poll)(struct psmouse *psmouse);

void (*pt_activate)(struct psmouse *psmouse);
void (*pt_deactivate)(struct psmouse *psmouse);
Index: work/drivers/input/mouse/psmouse-base.c
===================================================================
--- work.orig/drivers/input/mouse/psmouse-base.c
+++ work/drivers/input/mouse/psmouse-base.c
@@ -58,6 +58,10 @@ static unsigned int psmouse_resetafter;
module_param_named(resetafter, psmouse_resetafter, uint, 0644);
MODULE_PARM_DESC(resetafter, "Reset device after so many bad packets (0 = never).");

+static unsigned int psmouse_resync_time = 5;
+module_param_named(resync_time, psmouse_resync_time, uint, 0644);
+MODULE_PARM_DESC(resync_time, "How long can mouse stay idle before forcing resync (in seconds, 0 = never).");
+
PSMOUSE_DEFINE_ATTR(protocol, S_IWUSR | S_IRUGO,
NULL,
psmouse_attr_show_protocol, psmouse_attr_set_protocol);
@@ -70,12 +74,16 @@ PSMOUSE_DEFINE_ATTR(resolution, S_IWUSR
PSMOUSE_DEFINE_ATTR(resetafter, S_IWUSR | S_IRUGO,
(void *) offsetof(struct psmouse, resetafter),
psmouse_show_int_attr, psmouse_set_int_attr);
+PSMOUSE_DEFINE_ATTR(resync_time, S_IWUSR | S_IRUGO,
+ (void *) offsetof(struct psmouse, resync_time),
+ psmouse_show_int_attr, psmouse_set_int_attr);

static struct attribute *psmouse_attributes[] = {
&psmouse_attr_protocol.dattr.attr,
&psmouse_attr_rate.dattr.attr,
&psmouse_attr_resolution.dattr.attr,
&psmouse_attr_resetafter.dattr.attr,
+ &psmouse_attr_resync_time.dattr.attr,
NULL
};

@@ -98,6 +106,8 @@ __obsolete_setup("psmouse_rate=");
*/
static DECLARE_MUTEX(psmouse_sem);

+static struct workqueue_struct *kpsmoused_wq;
+
struct psmouse_protocol {
enum psmouse_type type;
char *name;
@@ -178,15 +188,79 @@ static psmouse_ret_t psmouse_process_byt
}

/*
- * psmouse_interrupt() handles incoming characters, either gathering them into
- * packets or passing them to the command routine as command output.
+ * __psmouse_set_state() sets new psmouse state and resets all flags.
+ */
+
+static inline void __psmouse_set_state(struct psmouse *psmouse, enum psmouse_state new_state)
+{
+ psmouse->state = new_state;
+ psmouse->pktcnt = psmouse->out_of_sync = 0;
+ psmouse->ps2dev.flags = 0;
+ psmouse->last = jiffies;
+}
+
+
+/*
+ * psmouse_set_state() sets new psmouse state and resets all flags and
+ * counters while holding serio lock so fighting with interrupt handler
+ * is not a concern.
+ */
+
+static void psmouse_set_state(struct psmouse *psmouse, enum psmouse_state new_state)
+{
+ serio_pause_rx(psmouse->ps2dev.serio);
+ __psmouse_set_state(psmouse, new_state);
+ serio_continue_rx(psmouse->ps2dev.serio);
+}
+
+/*
+ * psmouse_handle_byte() processes one byte of the input data stream
+ * by calling corresponding protocol handler.
+ */
+
+static int psmouse_handle_byte(struct psmouse *psmouse, struct pt_regs *regs)
+{
+ psmouse_ret_t rc = psmouse->protocol_handler(psmouse, regs);
+
+ switch (rc) {
+ case PSMOUSE_BAD_DATA:
+ if (psmouse->state == PSMOUSE_ACTIVATED) {
+ printk(KERN_WARNING "psmouse.c: %s at %s lost sync at byte %d\n",
+ psmouse->name, psmouse->phys, psmouse->pktcnt);
+ if (++psmouse->out_of_sync == psmouse->resetafter) {
+ __psmouse_set_state(psmouse, PSMOUSE_IGNORE);
+ printk(KERN_NOTICE "psmouse.c: issuing reconnect request\n");
+ serio_reconnect(psmouse->ps2dev.serio);
+ return -1;
+ }
+ }
+ psmouse->pktcnt = 0;
+ break;
+
+ case PSMOUSE_FULL_PACKET:
+ psmouse->pktcnt = 0;
+ if (psmouse->out_of_sync) {
+ psmouse->out_of_sync = 0;
+ printk(KERN_NOTICE "psmouse.c: %s at %s - driver resynched.\n",
+ psmouse->name, psmouse->phys);
+ }
+ break;
+
+ case PSMOUSE_GOOD_DATA:
+ break;
+ }
+ return 0;
+}
+
+/*
+ * psmouse_interrupt() handles incoming characters, either passing them
+ * for normal processing or gathering them as command response.
*/

static irqreturn_t psmouse_interrupt(struct serio *serio,
unsigned char data, unsigned int flags, struct pt_regs *regs)
{
struct psmouse *psmouse = serio_get_drvdata(serio);
- psmouse_ret_t rc;

if (psmouse->state == PSMOUSE_IGNORE)
goto out;
@@ -208,67 +282,58 @@ static irqreturn_t psmouse_interrupt(str
if (ps2_handle_response(&psmouse->ps2dev, data))
goto out;

- if (psmouse->state == PSMOUSE_INITIALIZING)
+ if (psmouse->state <= PSMOUSE_RESYNCING)
goto out;

if (psmouse->state == PSMOUSE_ACTIVATED &&
psmouse->pktcnt && time_after(jiffies, psmouse->last + HZ/2)) {
- printk(KERN_WARNING "psmouse.c: %s at %s lost synchronization, throwing %d bytes away.\n",
+ printk(KERN_INFO "psmouse.c: %s at %s lost synchronization, throwing %d bytes away.\n",
psmouse->name, psmouse->phys, psmouse->pktcnt);
- psmouse->pktcnt = 0;
+ psmouse->badbyte = psmouse->packet[0];
+ __psmouse_set_state(psmouse, PSMOUSE_RESYNCING);
+ queue_work(kpsmoused_wq, &psmouse->resync_work);
+ goto out;
}

- psmouse->last = jiffies;
psmouse->packet[psmouse->pktcnt++] = data;
-
- if (psmouse->packet[0] == PSMOUSE_RET_BAT) {
+/*
+ * Check if this is a new device announcement (0xAA 0x00)
+ */
+ if (unlikely(psmouse->packet[0] == PSMOUSE_RET_BAT && psmouse->pktcnt <= 2)) {
if (psmouse->pktcnt == 1)
goto out;

- if (psmouse->pktcnt == 2) {
- if (psmouse->packet[1] == PSMOUSE_RET_ID) {
- psmouse->state = PSMOUSE_IGNORE;
- serio_reconnect(serio);
- goto out;
- }
- if (psmouse->type == PSMOUSE_SYNAPTICS) {
- /* neither 0xAA nor 0x00 are valid first bytes
- * for a packet in absolute mode
- */
- psmouse->pktcnt = 0;
- goto out;
- }
+ if (psmouse->packet[1] == PSMOUSE_RET_ID) {
+ __psmouse_set_state(psmouse, PSMOUSE_IGNORE);
+ serio_reconnect(serio);
+ goto out;
}
- }
-
- rc = psmouse->protocol_handler(psmouse, regs);
+/*
+ * Not a new device, try processing first byte normally
+ */
+ psmouse->pktcnt = 1;
+ if (psmouse_handle_byte(psmouse, regs))
+ goto out;

- switch (rc) {
- case PSMOUSE_BAD_DATA:
- printk(KERN_WARNING "psmouse.c: %s at %s lost sync at byte %d\n",
- psmouse->name, psmouse->phys, psmouse->pktcnt);
- psmouse->pktcnt = 0;
+ psmouse->packet[psmouse->pktcnt++] = data;
+ }

- if (++psmouse->out_of_sync == psmouse->resetafter) {
- psmouse->state = PSMOUSE_IGNORE;
- printk(KERN_NOTICE "psmouse.c: issuing reconnect request\n");
- serio_reconnect(psmouse->ps2dev.serio);
- }
- break;
+/*
+ * See if we need to force resync because mouse was idle for too long
+ */
+ if (psmouse->state == PSMOUSE_ACTIVATED &&
+ psmouse->pktcnt == 1 && psmouse->resync_time &&
+ time_after(jiffies, psmouse->last + psmouse->resync_time * HZ)) {
+ psmouse->badbyte = psmouse->packet[0];
+ __psmouse_set_state(psmouse, PSMOUSE_RESYNCING);
+ queue_work(kpsmoused_wq, &psmouse->resync_work);
+ goto out;
+ }

- case PSMOUSE_FULL_PACKET:
- psmouse->pktcnt = 0;
- if (psmouse->out_of_sync) {
- psmouse->out_of_sync = 0;
- printk(KERN_NOTICE "psmouse.c: %s at %s - driver resynched.\n",
- psmouse->name, psmouse->phys);
- }
- break;
+ psmouse->last = jiffies;
+ psmouse_handle_byte(psmouse, regs);

- case PSMOUSE_GOOD_DATA:
- break;
- }
-out:
+ out:
return IRQ_HANDLED;
}

@@ -755,21 +820,6 @@ static void psmouse_initialize(struct ps
}

/*
- * psmouse_set_state() sets new psmouse state and resets all flags and
- * counters while holding serio lock so fighting with interrupt handler
- * is not a concern.
- */
-
-static void psmouse_set_state(struct psmouse *psmouse, enum psmouse_state new_state)
-{
- serio_pause_rx(psmouse->ps2dev.serio);
- psmouse->state = new_state;
- psmouse->pktcnt = psmouse->out_of_sync = 0;
- psmouse->ps2dev.flags = 0;
- serio_continue_rx(psmouse->ps2dev.serio);
-}
-
-/*
* psmouse_activate() enables the mouse so that we get motion reports from it.
*/

@@ -797,6 +847,100 @@ static void psmouse_deactivate(struct ps
psmouse_set_state(psmouse, PSMOUSE_CMD_MODE);
}

+/*
+ * psmouse_poll() - default poll hanlder. Everyone except for ALPS uses it.
+ */
+
+static int psmouse_poll(struct psmouse *psmouse)
+{
+ return ps2_command(&psmouse->ps2dev, psmouse->packet,
+ PSMOUSE_CMD_POLL | (psmouse->pktsize << 8));
+}
+
+
+/*
+ * psmouse_resync() attempts to re-validate current protocol.
+ */
+
+static void psmouse_resync(void *p)
+{
+ struct psmouse *psmouse = p, *parent = NULL;
+ struct serio *serio = psmouse->ps2dev.serio;
+ psmouse_ret_t rc = PSMOUSE_GOOD_DATA;
+ int failed = 0, enabled = 0;
+ int i;
+
+ down(&psmouse_sem);
+
+ if (psmouse->state != PSMOUSE_RESYNCING)
+ goto out;
+
+ if (serio->parent && serio->id.type == SERIO_PS_PSTHRU) {
+ parent = serio_get_drvdata(serio->parent);
+ psmouse_deactivate(parent);
+ }
+
+/*
+ * Some mice don't ACK commands sent while they are in the middle of
+ * transmitting motion packet. To avoid delay we use ps2_sendbyte()
+ * instead of ps2_command() which would wait for 200ms for an ACK
+ * that may never come.
+ */
+ ps2_sendbyte(&psmouse->ps2dev, PSMOUSE_CMD_DISABLE, 20);
+
+/*
+ * Poll the mouse. If it was reset the packet will be shorter than
+ * psmouse->pktsize and ps2_command will fail. We do not expect and
+ * do not handle scenario when mouse "upgrades" its protocol while
+ * disconnected since it would require additional delay. If we ever
+ * see a mouse that does it we'll adjust the code.
+ */
+ if (psmouse->poll(psmouse))
+ failed = 1;
+ else {
+ psmouse_set_state(psmouse, PSMOUSE_CMD_MODE);
+ for (i = 0; i < psmouse->pktsize; i++) {
+ psmouse->pktcnt++;
+ rc = psmouse->protocol_handler(psmouse, NULL);
+ if (rc != PSMOUSE_GOOD_DATA)
+ break;
+ }
+ if (rc != PSMOUSE_FULL_PACKET)
+ failed = 1;
+ psmouse_set_state(psmouse, PSMOUSE_RESYNCING);
+ }
+
+/*
+ * Now try to enable mouse. We try to do that even if poll failed and also
+ * repeat our attempts 5 times, otherwise we may be left out with disabled
+ * mouse.
+ */
+ for (i = 0; i < 5; i++) {
+ if (!ps2_command(&psmouse->ps2dev, NULL, PSMOUSE_CMD_ENABLE)) {
+ enabled = 1;
+ break;
+ }
+ msleep(200);
+ }
+
+ if (!enabled) {
+ printk(KERN_WARNING "psmouse.c: failed to re-enable mouse on %s\n",
+ psmouse->ps2dev.serio->phys);
+ failed = 1;
+ }
+
+ if (failed) {
+ psmouse_set_state(psmouse, PSMOUSE_IGNORE);
+ printk(KERN_INFO "psmouse.c: resync failed, issuing reconnect request\n");
+ serio_reconnect(serio);
+ } else
+ psmouse_set_state(psmouse, PSMOUSE_ACTIVATED);
+
+ if (parent)
+ psmouse_activate(parent);
+ out:
+ up(&psmouse_sem);
+}

/*
* psmouse_cleanup() resets the mouse into power-on state.
@@ -825,6 +969,11 @@ static void psmouse_disconnect(struct se

psmouse_set_state(psmouse, PSMOUSE_CMD_MODE);

+ /* make sure we don't have a resync in progress */
+ up(&psmouse_sem);
+ flush_workqueue(kpsmoused_wq);
+ down(&psmouse_sem);
+
if (serio->parent && serio->id.type == SERIO_PS_PSTHRU) {
parent = serio_get_drvdata(serio->parent);
psmouse_deactivate(parent);
@@ -862,6 +1011,7 @@ static int psmouse_switch_protocol(struc

psmouse->set_rate = psmouse_set_rate;
psmouse->set_resolution = psmouse_set_resolution;
+ psmouse->poll = psmouse_poll;
psmouse->protocol_handler = psmouse_process_byte;
psmouse->pktsize = 3;

@@ -917,6 +1067,7 @@ static int psmouse_connect(struct serio
goto out;

ps2_init(&psmouse->ps2dev, serio);
+ INIT_WORK(&psmouse->resync_work, psmouse_resync, psmouse);
psmouse->dev = input_dev;
sprintf(psmouse->phys, "%s/input0", serio->phys);

@@ -937,6 +1088,7 @@ static int psmouse_connect(struct serio
psmouse->rate = psmouse_rate;
psmouse->resolution = psmouse_resolution;
psmouse->resetafter = psmouse_resetafter;
+ psmouse->resync_time = parent ? 0 : psmouse_resync_time;
psmouse->smartscroll = psmouse_smartscroll;

psmouse_switch_protocol(psmouse, NULL);
@@ -1281,13 +1433,21 @@ static int psmouse_get_maxproto(char *bu

static int __init psmouse_init(void)
{
+ kpsmoused_wq = create_singlethread_workqueue("kpsmoused");
+ if (!kpsmoused_wq) {
+ printk(KERN_ERR "psmouse: failed to create kpsmoused workqueue\n");
+ return -ENOMEM;
+ }
+
serio_register_driver(&psmouse_drv);
+
return 0;
}

static void __exit psmouse_exit(void)
{
serio_unregister_driver(&psmouse_drv);
+ destroy_workqueue(kpsmoused_wq);
}

module_init(psmouse_init);
Index: work/drivers/input/mouse/logips2pp.c
===================================================================
--- work.orig/drivers/input/mouse/logips2pp.c
+++ work/drivers/input/mouse/logips2pp.c
@@ -117,7 +117,7 @@ static int ps2pp_cmd(struct psmouse *psm
if (psmouse_sliced_command(psmouse, command))
return -1;

- if (ps2_command(&psmouse->ps2dev, param, PSMOUSE_CMD_POLL))
+ if (ps2_command(&psmouse->ps2dev, param, PSMOUSE_CMD_POLL | 0x0300))
return -1;

return 0;
Index: work/drivers/input/mouse/alps.c
===================================================================
--- work.orig/drivers/input/mouse/alps.c
+++ work/drivers/input/mouse/alps.c
@@ -347,6 +347,39 @@ static int alps_tap_mode(struct psmouse
return 0;
}

+/*
+ * alps_poll() - poll the touchpad for current motion packet.
+ * Used in resync.
+ */
+static int alps_poll(struct psmouse *psmouse)
+{
+ struct alps_data *priv = psmouse->private;
+ unsigned char buf[6];
+
+ if (priv->i->flags & ALPS_PASS)
+ alps_passthrough_mode(psmouse, 1);
+
+ if (ps2_command(&psmouse->ps2dev, buf, PSMOUSE_CMD_POLL | (psmouse->pktsize << 8)))
+ return -1;
+
+ if ((buf[0] & priv->i->mask0) != priv->i->byte0)
+ return -1;
+
+ if (priv->i->flags & ALPS_PASS)
+ alps_passthrough_mode(psmouse, 0);
+
+ if ((psmouse->badbyte & 0xc8) == 0x08) {
+/*
+ * Poll the track stick ...
+ */
+ if (ps2_command(&psmouse->ps2dev, buf, PSMOUSE_CMD_POLL | (3 << 8)))
+ return -1;
+ }
+
+ memcpy(psmouse->packet, buf, sizeof(buf));
+ return 0;
+}
+
static int alps_reconnect(struct psmouse *psmouse)
{
struct alps_data *priv = psmouse->private;
@@ -450,10 +483,14 @@ int alps_init(struct psmouse *psmouse)
input_register_device(priv->dev2);

psmouse->protocol_handler = alps_process_byte;
+ psmouse->poll = alps_poll;
psmouse->disconnect = alps_disconnect;
psmouse->reconnect = alps_reconnect;
psmouse->pktsize = 6;

+ /* We are having trouble resyncing ALPS touchpads so disable it for now */
+ psmouse->resync_time = 0;
+
return 0;

init_fail:
Index: work/drivers/input/mouse/synaptics.c
===================================================================
--- work.orig/drivers/input/mouse/synaptics.c
+++ work/drivers/input/mouse/synaptics.c
@@ -652,6 +652,8 @@ int synaptics_init(struct psmouse *psmou
psmouse->disconnect = synaptics_disconnect;
psmouse->reconnect = synaptics_reconnect;
psmouse->pktsize = 6;
+ /* Synaptics can usually stay in sync without extra help */
+ psmouse->resync_time = 0;

if (SYN_CAP_PASS_THROUGH(priv->capabilities))
synaptics_pt_create(psmouse);

2005-11-20 18:38:22

[permalink] [raw]

Subject: Re: 2.6.15-rc1-mm2 unsusable on DELL Inspiron 8200, 2.6.15-rc1 works fine

* Dmitry Torokhov <[email protected]> [2005-11-18 22:07:19 -0500]:

> On Friday 18 November 2005 13:29, Marc Koschewski wrote:
> > Nov 18 12:58:37 stiffy kernel: psmouse.c: Wheel Mouse at isa0060/serio1/input0 lost synchronization, throwing 1 bytes away.
> >
> > SOME STUFF MISSING? HUH?
> >
> > Nov 18 13:03:14 stiffy kernel: psmouse.c: resync failed, issuing reconnect request
> >
>
> Hm, this worries me a bit... Could you please try appying the patch
> below to plain 2.6.15-rc1 and see if mouse starts misbehaving again?

Dmitry,

I applied the 5 patches to a plain 2.6.15-rc1. The mouse was well as if it was
in an unpatched kernel. The problem just occured in 2.6.15-rc1-mmX.
Plain 2.6.15-rc1 was fine before as well. So: actually no change.

Need any more info?

Marc

2005-11-22 03:43:53

[permalink] [raw]

Subject: Re: 2.6.15-rc1-mm2 unsusable on DELL Inspiron 8200, 2.6.15-rc1 works fine

On Sunday 20 November 2005 12:14, Marc Koschewski wrote:
> * Dmitry Torokhov <[email protected]> [2005-11-18 22:07:19 -0500]:
>
> > On Friday 18 November 2005 13:29, Marc Koschewski wrote:
> > > Nov 18 12:58:37 stiffy kernel: psmouse.c: Wheel Mouse at isa0060/serio1/input0 lost synchronization, throwing 1 bytes away.
> > >
> > > SOME STUFF MISSING? HUH?
> > >
> > > Nov 18 13:03:14 stiffy kernel: psmouse.c: resync failed, issuing reconnect request
> > >
> >
> > Hm, this worries me a bit... Could you please try appying the patch
> > below to plain 2.6.15-rc1 and see if mouse starts misbehaving again?
>
> Dmitry,
>
> I applied the 5 patches to a plain 2.6.15-rc1. The mouse was well as if it was
> in an unpatched kernel. The problem just occured in 2.6.15-rc1-mmX.
> Plain 2.6.15-rc1 was fine before as well. So: actually no change.
>
> Need any more info?
>

Marc,

Thank you for testing the patch. It proves that your mouse troubles
were not caused by the patch I made so I am very happy. "No change"
is the result I wanted to hear ;)

--
Dmitry

2005-11-22 10:14:12

[permalink] [raw]

Subject: Re: 2.6.15-rc1-mm2 unsusable on DELL Inspiron 8200, 2.6.15-rc1 works fine

* Dmitry Torokhov <[email protected]> [2005-11-21 22:43:50 -0500]:

> On Sunday 20 November 2005 12:14, Marc Koschewski wrote:
> > * Dmitry Torokhov <[email protected]> [2005-11-18 22:07:19 -0500]:
> >
> > > On Friday 18 November 2005 13:29, Marc Koschewski wrote:
> > > > Nov 18 12:58:37 stiffy kernel: psmouse.c: Wheel Mouse at isa0060/serio1/input0 lost synchronization, throwing 1 bytes away.
> > > >
> > > > SOME STUFF MISSING? HUH?
> > > >
> > > > Nov 18 13:03:14 stiffy kernel: psmouse.c: resync failed, issuing reconnect request
> > > >
> > >
> > > Hm, this worries me a bit... Could you please try appying the patch
> > > below to plain 2.6.15-rc1 and see if mouse starts misbehaving again?
> >
> > Dmitry,
> >
> > I applied the 5 patches to a plain 2.6.15-rc1. The mouse was well as if it was
> > in an unpatched kernel. The problem just occured in 2.6.15-rc1-mmX.
> > Plain 2.6.15-rc1 was fine before as well. So: actually no change.
> >
> > Need any more info?
> >
>
> Marc,
>
> Thank you for testing the patch. It proves that your mouse troubles
> were not caused by the patch I made so I am very happy. "No change"
> is the result I wanted to hear ;)

Hey,

ok, I just thought you wanted me to say that. ;) But what's the problem
then? I didn't change any configs (expect the 'udev is no longer loading
mousedev'-problem on Debian - I put mousedev into /etc/modules).
However, this doesn't affect the mouse in 2.6.15-rc1 and 2.6.15-rc2. Is
this a timing issue? udev-detected modules get loaded earlier than
/etc/modules modules get loaded in Debian, don't they?. Maybe some resource
mess?

Marc

2005-11-23 19:56:46

[permalink] [raw]

Subject: psmouse unusable in -mm series (was: 2.6.15-rc1-mm2 unsusable on DELL Inspiron 8200, 2.6.15-rc1 works fine)

* Dmitry Torokhov <[email protected]> [2005-11-21 22:43:50 -0500]:

> On Sunday 20 November 2005 12:14, Marc Koschewski wrote:
> > * Dmitry Torokhov <[email protected]> [2005-11-18 22:07:19 -0500]:
> >
> > > On Friday 18 November 2005 13:29, Marc Koschewski wrote:
> > > > Nov 18 12:58:37 stiffy kernel: psmouse.c: Wheel Mouse at isa0060/serio1/input0 lost synchronization, throwing 1 bytes away.
> > > >
> > > > SOME STUFF MISSING? HUH?
> > > >
> > > > Nov 18 13:03:14 stiffy kernel: psmouse.c: resync failed, issuing reconnect request
> > > >
> > >
> > > Hm, this worries me a bit... Could you please try appying the patch
> > > below to plain 2.6.15-rc1 and see if mouse starts misbehaving again?
> >
> > Dmitry,
> >
> > I applied the 5 patches to a plain 2.6.15-rc1. The mouse was well as if it was
> > in an unpatched kernel. The problem just occured in 2.6.15-rc1-mmX.
> > Plain 2.6.15-rc1 was fine before as well. So: actually no change.
> >
> > Need any more info?
> >
>
> Marc,
>
> Thank you for testing the patch. It proves that your mouse troubles
> were not caused by the patch I made so I am very happy. "No change"
> is the result I wanted to hear ;)
>

Dmitry,

there's a bug report filed against Debian's udev. You can read it here:

http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=340202

The bug report, however, states that the problem is caused by udev under
all variants of kernel 2.6.15. I'm writing this mail while running
2.6.15-rc1 and the mouse definitely works. Do you have any other hint?
Seems to me like the bug report is only half the truth...

Regards,
Marc

2005-11-24 02:29:07

by Ed Tomlinson

[permalink] [raw]

Subject: Re: psmouse unusable in -mm series (was: 2.6.15-rc1-mm2 unsusable on DELL Inspiron 8200, 2.6.15-rc1 works fine)

On Wednesday 23 November 2005 14:57, Marc Koschewski wrote:
> * Dmitry Torokhov <[email protected]> [2005-11-21 22:43:50 -0500]:
>
> > On Sunday 20 November 2005 12:14, Marc Koschewski wrote:
> > > * Dmitry Torokhov <[email protected]> [2005-11-18 22:07:19 -0500]:
> > >
> > > > On Friday 18 November 2005 13:29, Marc Koschewski wrote:
> > > > > Nov 18 12:58:37 stiffy kernel: psmouse.c: Wheel Mouse at isa0060/serio1/input0 lost synchronization, throwing 1 bytes away.
> > > > >
> > > > > SOME STUFF MISSING? HUH?
> > > > >
> > > > > Nov 18 13:03:14 stiffy kernel: psmouse.c: resync failed, issuing reconnect request
> > > > >
> > > >
> > > > Hm, this worries me a bit... Could you please try appying the patch
> > > > below to plain 2.6.15-rc1 and see if mouse starts misbehaving again?
> > >
> > > Dmitry,
> > >
> > > I applied the 5 patches to a plain 2.6.15-rc1. The mouse was well as if it was
> > > in an unpatched kernel. The problem just occured in 2.6.15-rc1-mmX.
> > > Plain 2.6.15-rc1 was fine before as well. So: actually no change.
> > >
> > > Need any more info?
> > >
> >
> > Marc,
> >
> > Thank you for testing the patch. It proves that your mouse troubles
> > were not caused by the patch I made so I am very happy. "No change"
> > is the result I wanted to hear ;)
> >
>
> Dmitry,
>
> there's a bug report filed against Debian's udev. You can read it here:
>
> http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=340202
>
> The bug report, however, states that the problem is caused by udev under
> all variants of kernel 2.6.15. I'm writing this mail while running
> 2.6.15-rc1 and the mouse definitely works. Do you have any other hint?
> Seems to me like the bug report is only half the truth...

Marc,

Are you, by some slim chance, manually loading mousedev ( via /etc/modules) or
an init script? If so your mouse will work.

Ed Tomlinson

2005-11-24 04:41:24

[permalink] [raw]

Subject: Re: psmouse unusable in -mm series (was: 2.6.15-rc1-mm2 unsusable on DELL Inspiron 8200, 2.6.15-rc1 works fine)

On Wednesday 23 November 2005 21:29, Ed Tomlinson wrote:
> On Wednesday 23 November 2005 14:57, Marc Koschewski wrote:
> > * Dmitry Torokhov <[email protected]> [2005-11-21 22:43:50 -0500]:
> >
> > > On Sunday 20 November 2005 12:14, Marc Koschewski wrote:
> > > > * Dmitry Torokhov <[email protected]> [2005-11-18 22:07:19 -0500]:
> > > >
> > > > > On Friday 18 November 2005 13:29, Marc Koschewski wrote:
> > > > > > Nov 18 12:58:37 stiffy kernel: psmouse.c: Wheel Mouse at isa0060/serio1/input0 lost synchronization, throwing 1 bytes away.
> > > > > >
> > > > > > SOME STUFF MISSING? HUH?
> > > > > >
> > > > > > Nov 18 13:03:14 stiffy kernel: psmouse.c: resync failed, issuing reconnect request
> > > > > >
> > > > >
> > > > > Hm, this worries me a bit... Could you please try appying the patch
> > > > > below to plain 2.6.15-rc1 and see if mouse starts misbehaving again?
> > > >
> > > > Dmitry,
> > > >
> > > > I applied the 5 patches to a plain 2.6.15-rc1. The mouse was well as if it was
> > > > in an unpatched kernel. The problem just occured in 2.6.15-rc1-mmX.
> > > > Plain 2.6.15-rc1 was fine before as well. So: actually no change.
> > > >
> > > > Need any more info?
> > > >
> > >
> > > Marc,
> > >
> > > Thank you for testing the patch. It proves that your mouse troubles
> > > were not caused by the patch I made so I am very happy. "No change"
> > > is the result I wanted to hear ;)
> > >
> >
> > Dmitry,
> >
> > there's a bug report filed against Debian's udev. You can read it here:
> >
> > http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=340202
> >
> > The bug report, however, states that the problem is caused by udev under
> > all variants of kernel 2.6.15. I'm writing this mail while running
> > 2.6.15-rc1 and the mouse definitely works. Do you have any other hint?
> > Seems to me like the bug report is only half the truth...
>
> Marc,
>
> Are you, by some slim chance, manually loading mousedev ( via /etc/modules) or
> an init script? If so your mouse will work.
>

The only change in input between 2.6.15-rc1 and 2.6.15-rc-mm2 is that
psmouse resync patch. Marc, is you revert psmouse resync patch from -mm
does the mouzse still misbehaves?

--
Dmitry

2005-11-24 08:41:04

[permalink] [raw]

Subject: Re: psmouse unusable in -mm series (was: 2.6.15-rc1-mm2 unsusable on DELL Inspiron 8200, 2.6.15-rc1 works fine)

* Ed Tomlinson <[email protected]> [2005-11-23 21:29:35 -0500]:

> On Wednesday 23 November 2005 14:57, Marc Koschewski wrote:
> > * Dmitry Torokhov <[email protected]> [2005-11-21 22:43:50 -0500]:
> >
> > > On Sunday 20 November 2005 12:14, Marc Koschewski wrote:
> > > > * Dmitry Torokhov <[email protected]> [2005-11-18 22:07:19 -0500]:
> > > >
> > > > > On Friday 18 November 2005 13:29, Marc Koschewski wrote:
> > > > > > Nov 18 12:58:37 stiffy kernel: psmouse.c: Wheel Mouse at isa0060/serio1/input0 lost synchronization, throwing 1 bytes away.
> > > > > >
> > > > > > SOME STUFF MISSING? HUH?
> > > > > >
> > > > > > Nov 18 13:03:14 stiffy kernel: psmouse.c: resync failed, issuing reconnect request
> > > > > >
> > > > >
> > > > > Hm, this worries me a bit... Could you please try appying the patch
> > > > > below to plain 2.6.15-rc1 and see if mouse starts misbehaving again?
> > > >
> > > > Dmitry,
> > > >
> > > > I applied the 5 patches to a plain 2.6.15-rc1. The mouse was well as if it was
> > > > in an unpatched kernel. The problem just occured in 2.6.15-rc1-mmX.
> > > > Plain 2.6.15-rc1 was fine before as well. So: actually no change.
> > > >
> > > > Need any more info?
> > > >
> > >
> > > Marc,
> > >
> > > Thank you for testing the patch. It proves that your mouse troubles
> > > were not caused by the patch I made so I am very happy. "No change"
> > > is the result I wanted to hear ;)
> > >
> >
> > Dmitry,
> >
> > there's a bug report filed against Debian's udev. You can read it here:
> >
> > http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=340202
> >
> > The bug report, however, states that the problem is caused by udev under
> > all variants of kernel 2.6.15. I'm writing this mail while running
> > 2.6.15-rc1 and the mouse definitely works. Do you have any other hint?
> > Seems to me like the bug report is only half the truth...
>
> Marc,
>
> Are you, by some slim chance, manually loading mousedev ( via /etc/modules) or
> an init script? If so your mouse will work.
>

Hi Ed,

yes, I do load mousedev as well as psmouse in /etc/modules due to udev
no longer loading it. my problem, however, is _not_ that teh module
doesn't get loaded. It's the mouse pointer jumping all over the screen.
After some time (inactivity?) it calms down and is usable. Then, after
some minutes it again jumps wild all over the place, seems to click
stuff on it's own, and again calms down, ...

This problem shouldn't be udev related.

One note: the pointer seems to be in love with the corners of the
screen. It always goes there when it goes wild... ;)

Marc

2005-11-24 12:16:38

by Ed Tomlinson

[permalink] [raw]

Subject: Re: psmouse unusable in -mm series (was: 2.6.15-rc1-mm2 unsusable on DELL Inspiron 8200, 2.6.15-rc1 works fine)

On Thursday 24 November 2005 03:41, Marc Koschewski wrote:
> * Ed Tomlinson <[email protected]> [2005-11-23 21:29:35 -0500]:
>
> > On Wednesday 23 November 2005 14:57, Marc Koschewski wrote:
> > > * Dmitry Torokhov <[email protected]> [2005-11-21 22:43:50 -0500]:
> > >
> > > > On Sunday 20 November 2005 12:14, Marc Koschewski wrote:
> > > > > * Dmitry Torokhov <[email protected]> [2005-11-18 22:07:19 -0500]:
> > > > >
> > > > > > On Friday 18 November 2005 13:29, Marc Koschewski wrote:
> > > > > > > Nov 18 12:58:37 stiffy kernel: psmouse.c: Wheel Mouse at isa0060/serio1/input0 lost synchronization, throwing 1 bytes away.
> > > > > > >
> > > > > > > SOME STUFF MISSING? HUH?
> > > > > > >
> > > > > > > Nov 18 13:03:14 stiffy kernel: psmouse.c: resync failed, issuing reconnect request
> > > > > > >
> > > > > >
> > > > > > Hm, this worries me a bit... Could you please try appying the patch
> > > > > > below to plain 2.6.15-rc1 and see if mouse starts misbehaving again?
> > > > >
> > > > > Dmitry,
> > > > >
> > > > > I applied the 5 patches to a plain 2.6.15-rc1. The mouse was well as if it was
> > > > > in an unpatched kernel. The problem just occured in 2.6.15-rc1-mmX.
> > > > > Plain 2.6.15-rc1 was fine before as well. So: actually no change.
> > > > >
> > > > > Need any more info?
> > > > >
> > > >
> > > > Marc,
> > > >
> > > > Thank you for testing the patch. It proves that your mouse troubles
> > > > were not caused by the patch I made so I am very happy. "No change"
> > > > is the result I wanted to hear ;)
> > > >
> > >
> > > Dmitry,
> > >
> > > there's a bug report filed against Debian's udev. You can read it here:
> > >
> > > http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=340202
> > >
> > > The bug report, however, states that the problem is caused by udev under
> > > all variants of kernel 2.6.15. I'm writing this mail while running
> > > 2.6.15-rc1 and the mouse definitely works. Do you have any other hint?
> > > Seems to me like the bug report is only half the truth...
> >
> > Marc,
> >
> > Are you, by some slim chance, manually loading mousedev ( via /etc/modules) or
> > an init script? If so your mouse will work.
> >
>
> Hi Ed,
>
> yes, I do load mousedev as well as psmouse in /etc/modules due to udev
> no longer loading it. my problem, however, is _not_ that teh module
> doesn't get loaded. It's the mouse pointer jumping all over the screen.
> After some time (inactivity?) it calms down and is usable. Then, after
> some minutes it again jumps wild all over the place, seems to click
> stuff on it's own, and again calms down, ...
>
> This problem shouldn't be udev related.
>
> One note: the pointer seems to be in love with the corners of the
> screen. It always goes there when it goes wild... ;)

Marc,

It looks like you are seeing a different bug. The one opened for debian user space
covers mousedev not being loaded if the kernel is 2.6.15, which leads to no /dev/input

Ed Tomlinson

2005-11-24 12:44:27

[permalink] [raw]

Subject: Re: psmouse unusable in -mm series (was: 2.6.15-rc1-mm2 unsusable on DELL Inspiron 8200, 2.6.15-rc1 works fine)

* Ed Tomlinson <[email protected]> [2005-11-24 07:17:10 -0500]:

> On Thursday 24 November 2005 03:41, Marc Koschewski wrote:
> > * Ed Tomlinson <[email protected]> [2005-11-23 21:29:35 -0500]:
> >
> > > On Wednesday 23 November 2005 14:57, Marc Koschewski wrote:
> > > > * Dmitry Torokhov <[email protected]> [2005-11-21 22:43:50 -0500]:
> > > >
> > > > > On Sunday 20 November 2005 12:14, Marc Koschewski wrote:
> > > > > > * Dmitry Torokhov <[email protected]> [2005-11-18 22:07:19 -0500]:
> > > > > >
> > > > > > > On Friday 18 November 2005 13:29, Marc Koschewski wrote:
> > > > > > > > Nov 18 12:58:37 stiffy kernel: psmouse.c: Wheel Mouse at isa0060/serio1/input0 lost synchronization, throwing 1 bytes away.
> > > > > > > >
> > > > > > > > SOME STUFF MISSING? HUH?
> > > > > > > >
> > > > > > > > Nov 18 13:03:14 stiffy kernel: psmouse.c: resync failed, issuing reconnect request
> > > > > > > >
> > > > > > >
> > > > > > > Hm, this worries me a bit... Could you please try appying the patch
> > > > > > > below to plain 2.6.15-rc1 and see if mouse starts misbehaving again?
> > > > > >
> > > > > > Dmitry,
> > > > > >
> > > > > > I applied the 5 patches to a plain 2.6.15-rc1. The mouse was well as if it was
> > > > > > in an unpatched kernel. The problem just occured in 2.6.15-rc1-mmX.
> > > > > > Plain 2.6.15-rc1 was fine before as well. So: actually no change.
> > > > > >
> > > > > > Need any more info?
> > > > > >
> > > > >
> > > > > Marc,
> > > > >
> > > > > Thank you for testing the patch. It proves that your mouse troubles
> > > > > were not caused by the patch I made so I am very happy. "No change"
> > > > > is the result I wanted to hear ;)
> > > > >
> > > >
> > > > Dmitry,
> > > >
> > > > there's a bug report filed against Debian's udev. You can read it here:
> > > >
> > > > http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=340202
> > > >
> > > > The bug report, however, states that the problem is caused by udev under
> > > > all variants of kernel 2.6.15. I'm writing this mail while running
> > > > 2.6.15-rc1 and the mouse definitely works. Do you have any other hint?
> > > > Seems to me like the bug report is only half the truth...
> > >
> > > Marc,
> > >
> > > Are you, by some slim chance, manually loading mousedev ( via /etc/modules) or
> > > an init script? If so your mouse will work.
> > >
> >
> > Hi Ed,
> >
> > yes, I do load mousedev as well as psmouse in /etc/modules due to udev
> > no longer loading it. my problem, however, is _not_ that teh module
> > doesn't get loaded. It's the mouse pointer jumping all over the screen.
> > After some time (inactivity?) it calms down and is usable. Then, after
> > some minutes it again jumps wild all over the place, seems to click
> > stuff on it's own, and again calms down, ...
> >
> > This problem shouldn't be udev related.
> >
> > One note: the pointer seems to be in love with the corners of the
> > screen. It always goes there when it goes wild... ;)
>
> Marc,
>
> It looks like you are seeing a different bug. The one opened for debian user space
> covers mousedev not being loaded if the kernel is 2.6.15, which leads to no /dev/input
>

That's what I think, thus the report on LKLM. But noone but me seems to
be trapped into it until... :/

Marc

2005-11-24 20:23:08

[permalink] [raw]

Subject: Re: psmouse unusable in -mm series (was: 2.6.15-rc1-mm2 unsusable on DELL Inspiron 8200, 2.6.15-rc1 works fine)

On Thursday, 24 of November 2005 13:44, Marc Koschewski wrote:
}-- snip --{
> > It looks like you are seeing a different bug. The one opened for debian user space
> > covers mousedev not being loaded if the kernel is 2.6.15, which leads to no /dev/input
> >
>
> That's what I think, thus the report on LKLM. But noone but me seems to
> be trapped into it until... :/

FWIW, my touchpad doesn't work with -rc2-mm1 too (usually I use a USB mouse,
so I didn't notice before). Here's what dmesg says about it:

Synaptics Touchpad, model: 1, fw: 5.9, id: 0x926eb1, caps: 0x804719/0x0
input: SynPS/2 Synaptics TouchPad as /class/input/input2

The box is an Asus L5D (x86-64).

Greetings,
Rafael

2005-11-24 21:19:29

[permalink] [raw]

Subject: Re: psmouse unusable in -mm series (was: 2.6.15-rc1-mm2 unsusable on DELL Inspiron 8200, 2.6.15-rc1 works fine)

Update:

On Thursday, 24 of November 2005 21:23, Rafael J. Wysocki wrote:
> On Thursday, 24 of November 2005 13:44, Marc Koschewski wrote:
> }-- snip --{
> > > It looks like you are seeing a different bug. The one opened for debian user space
> > > covers mousedev not being loaded if the kernel is 2.6.15, which leads to no /dev/input
> > >
> >
> > That's what I think, thus the report on LKLM. But noone but me seems to
> > be trapped into it until... :/
>
> FWIW, my touchpad doesn't work with -rc2-mm1 too (usually I use a USB mouse,
> so I didn't notice before). Here's what dmesg says about it:
>
> Synaptics Touchpad, model: 1, fw: 5.9, id: 0x926eb1, caps: 0x804719/0x0
> input: SynPS/2 Synaptics TouchPad as /class/input/input2
>
> The box is an Asus L5D (x86-64).

Actually, it works on the console (ie with gpm), but X is unable to use it,
apparently. However it used to be, at least on 2.6.14-git9 (this is the latest
non-mm kernel I've been able to test quickly on this box).

Marc, does your touchpad work with gpm?

Rafael

2005-11-25 08:22:55

by Vojtech Pavlik

[permalink] [raw]

Subject: Re: psmouse unusable in -mm series (was: 2.6.15-rc1-mm2 unsusable on DELL Inspiron 8200, 2.6.15-rc1 works fine)

On Thu, Nov 24, 2005 at 10:20:25PM +0100, Rafael J. Wysocki wrote:
> Update:
>
> On Thursday, 24 of November 2005 21:23, Rafael J. Wysocki wrote:
> > On Thursday, 24 of November 2005 13:44, Marc Koschewski wrote:
> > }-- snip --{
> > > > It looks like you are seeing a different bug. The one opened for debian user space
> > > > covers mousedev not being loaded if the kernel is 2.6.15, which leads to no /dev/input
> > > >
> > >
> > > That's what I think, thus the report on LKLM. But noone but me seems to
> > > be trapped into it until... :/
> >
> > FWIW, my touchpad doesn't work with -rc2-mm1 too (usually I use a USB mouse,
> > so I didn't notice before). Here's what dmesg says about it:
> >
> > Synaptics Touchpad, model: 1, fw: 5.9, id: 0x926eb1, caps: 0x804719/0x0
> > input: SynPS/2 Synaptics TouchPad as /class/input/input2
> >
> > The box is an Asus L5D (x86-64).
>
> Actually, it works on the console (ie with gpm), but X is unable to use it,
> apparently. However it used to be, at least on 2.6.14-git9 (this is the latest
> non-mm kernel I've been able to test quickly on this box).
>
> Marc, does your touchpad work with gpm?

What's in your relevant xorg.conf sections?

--
Vojtech Pavlik
SuSE Labs, SuSE CR

2005-11-25 13:50:23

[permalink] [raw]

Subject: Re: psmouse unusable in -mm series (was: 2.6.15-rc1-mm2 unsusable on DELL Inspiron 8200, 2.6.15-rc1 works fine)

* Vojtech Pavlik <[email protected]> [2005-11-25 09:22:53 +0100]:

> On Thu, Nov 24, 2005 at 10:20:25PM +0100, Rafael J. Wysocki wrote:
> > Update:
> >
> > On Thursday, 24 of November 2005 21:23, Rafael J. Wysocki wrote:
> > > On Thursday, 24 of November 2005 13:44, Marc Koschewski wrote:
> > > }-- snip --{
> > > > > It looks like you are seeing a different bug. The one opened for debian user space
> > > > > covers mousedev not being loaded if the kernel is 2.6.15, which leads to no /dev/input
> > > > >
> > > >
> > > > That's what I think, thus the report on LKLM. But noone but me seems to
> > > > be trapped into it until... :/
> > >
> > > FWIW, my touchpad doesn't work with -rc2-mm1 too (usually I use a USB mouse,
> > > so I didn't notice before). Here's what dmesg says about it:
> > >
> > > Synaptics Touchpad, model: 1, fw: 5.9, id: 0x926eb1, caps: 0x804719/0x0
> > > input: SynPS/2 Synaptics TouchPad as /class/input/input2
> > >
> > > The box is an Asus L5D (x86-64).
> >
> > Actually, it works on the console (ie with gpm), but X is unable to use it,
> > apparently. However it used to be, at least on 2.6.14-git9 (this is the latest
> > non-mm kernel I've been able to test quickly on this box).
> >
> > Marc, does your touchpad work with gpm?
>
> What's in your relevant xorg.conf sections?

Section "InputDevice"
Identifier "id_mouse"
Driver "mouse"
Option "CorePointer"
Option "Device" "/dev/input/mice"
Option "Protocol" "ImPS/2"
Option "ZAxisMapping" "4 5"
EndSection

Section "ServerLayout"
Identifier "home"
Screen 0 "id_screen_lcd" 0 0
InputDevice "id_mouse" "CorePointer"
InputDevice "id_keyboard" "CoreKeyboard"
#Option "Xinerama" "on"
Option "Xinerama" "off"
EndSection

Section "ServerFlags"
Option "AllowMouseOpenFail" "true"
Option "DontZap" "true"
Option "DontVTSwitch" "false"
Option "DefaultServerLayout" "home"
EndSection

2005-11-25 13:53:25

[permalink] [raw]

Subject: Re: psmouse unusable in -mm series (was: 2.6.15-rc1-mm2 unsusable on DELL Inspiron 8200, 2.6.15-rc1 works fine)

* Rafael J. Wysocki <[email protected]> [2005-11-24 21:23:59 +0100]:

> On Thursday, 24 of November 2005 13:44, Marc Koschewski wrote:
> }-- snip --{
> > > It looks like you are seeing a different bug. The one opened for debian user space
> > > covers mousedev not being loaded if the kernel is 2.6.15, which leads to no /dev/input
> > >
> >
> > That's what I think, thus the report on LKLM. But noone but me seems to
> > be trapped into it until... :/
>
> FWIW, my touchpad doesn't work with -rc2-mm1 too (usually I use a USB mouse,
> so I didn't notice before). Here's what dmesg says about it:
>
> Synaptics Touchpad, model: 1, fw: 5.9, id: 0x926eb1, caps: 0x804719/0x0
> input: SynPS/2 Synaptics TouchPad as /class/input/input2
>
> The box is an Asus L5D (x86-64).

That's what 2.6.15-rc2 reports for my input devices:

stiffy:/var/cache/apt/archives# dmesg | grep input
input: AT Translated Set 2 keyboard as /class/input/input0
input: PC Speaker as /class/input/input1
input: ImPS/2 Generic Wheel Mouse as /class/input/input2
drivers/usb/input/usbmouse.c: v1.6:USB HID Boot Protocol mouse driver

Regards,
Marc

2005-11-25 21:15:28

[permalink] [raw]

Subject: Re: psmouse unusable in -mm series (was: 2.6.15-rc1-mm2 unsusable on DELL Inspiron 8200, 2.6.15-rc1 works fine)

On Friday, 25 of November 2005 09:22, Vojtech Pavlik wrote:
> On Thu, Nov 24, 2005 at 10:20:25PM +0100, Rafael J. Wysocki wrote:
> > Update:
> >
> > On Thursday, 24 of November 2005 21:23, Rafael J. Wysocki wrote:
> > > On Thursday, 24 of November 2005 13:44, Marc Koschewski wrote:
> > > }-- snip --{
> > > > > It looks like you are seeing a different bug. The one opened for debian user space
> > > > > covers mousedev not being loaded if the kernel is 2.6.15, which leads to no /dev/input
> > > > >
> > > >
> > > > That's what I think, thus the report on LKLM. But noone but me seems to
> > > > be trapped into it until... :/
> > >
> > > FWIW, my touchpad doesn't work with -rc2-mm1 too (usually I use a USB mouse,
> > > so I didn't notice before). Here's what dmesg says about it:
> > >
> > > Synaptics Touchpad, model: 1, fw: 5.9, id: 0x926eb1, caps: 0x804719/0x0
> > > input: SynPS/2 Synaptics TouchPad as /class/input/input2
> > >
> > > The box is an Asus L5D (x86-64).
> >
> > Actually, it works on the console (ie with gpm), but X is unable to use it,
> > apparently. However it used to be, at least on 2.6.14-git9 (this is the latest
> > non-mm kernel I've been able to test quickly on this box).
> >
> > Marc, does your touchpad work with gpm?
>
> What's in your relevant xorg.conf sections?

Section "ServerFlags"
Option "AllowMouseOpenFail"
EndSection

Section "InputDevice"
Driver "mouse"
Identifier "Mouse[1]"
Option "Device" "/dev/input/mice"
Option "Name" "Logitech Optical USB Mouse"
Option "Protocol" "explorerps/2"
Option "Vendor" "Sysp"
Option "ZAxisMapping" "4 5"
EndSection

Section "InputDevice"
Driver "synaptics"
Identifier "Mouse[3]"
Option "Device" "/dev/input/mice"
Option "Emulate3Buttons" "on"
Option "InputFashion" "Mouse"
Option "Name" "Synaptics;Touchpad"
Option "Protocol" "explorerps/2"
Option "SHMConfig" "on"
Option "Vendor" "Sysp"
Option "ZAxisMapping" "4 5"
EndSection

Section "ServerLayout"
Identifier "Layout[all]"
InputDevice "Keyboard[0]" "CoreKeyboard"
InputDevice "Mouse[1]" "CorePointer"
InputDevice "Mouse[3]" "SendCoreEvents"
Option "Clone" "off"
Option "Xinerama" "off"
Screen "Screen[0]"
EndSection

2005-11-25 21:41:31

[permalink] [raw]

Subject: Re: psmouse unusable in -mm series (was: 2.6.15-rc1-mm2 unsusable on DELL Inspiron 8200, 2.6.15-rc1 works fine)

On Friday 25 November 2005 16:16, Rafael J. Wysocki wrote:
> On Friday, 25 of November 2005 09:22, Vojtech Pavlik wrote:
> > On Thu, Nov 24, 2005 at 10:20:25PM +0100, Rafael J. Wysocki wrote:
> > > Update:
> > >
> > > On Thursday, 24 of November 2005 21:23, Rafael J. Wysocki wrote:
> > > > On Thursday, 24 of November 2005 13:44, Marc Koschewski wrote:
> > > > }-- snip --{
> > > > > > It looks like you are seeing a different bug. The one opened for debian user space
> > > > > > covers mousedev not being loaded if the kernel is 2.6.15, which leads to no /dev/input
> > > > > >
> > > > >
> > > > > That's what I think, thus the report on LKLM. But noone but me seems to
> > > > > be trapped into it until... :/
> > > >
> > > > FWIW, my touchpad doesn't work with -rc2-mm1 too (usually I use a USB mouse,
> > > > so I didn't notice before). Here's what dmesg says about it:
> > > >
> > > > Synaptics Touchpad, model: 1, fw: 5.9, id: 0x926eb1, caps: 0x804719/0x0
> > > > input: SynPS/2 Synaptics TouchPad as /class/input/input2
> > > >
> > > > The box is an Asus L5D (x86-64).
> > >
> > > Actually, it works on the console (ie with gpm), but X is unable to use it,
> > > apparently. However it used to be, at least on 2.6.14-git9 (this is the latest
> > > non-mm kernel I've been able to test quickly on this box).
> > >
> > > Marc, does your touchpad work with gpm?
> >
> > What's in your relevant xorg.conf sections?
>
> Section "ServerFlags"
> Option "AllowMouseOpenFail"
> EndSection
>
> Section "InputDevice"
> Driver "mouse"
> Identifier "Mouse[1]"
> Option "Device" "/dev/input/mice"
> Option "Name" "Logitech Optical USB Mouse"
> Option "Protocol" "explorerps/2"
> Option "Vendor" "Sysp"
> Option "ZAxisMapping" "4 5"
> EndSection
>
> Section "InputDevice"
> Driver "synaptics"
> Identifier "Mouse[3]"
> Option "Device" "/dev/input/mice"
> Option "Emulate3Buttons" "on"
> Option "InputFashion" "Mouse"
> Option "Name" "Synaptics;Touchpad"
> Option "Protocol" "explorerps/2"
> Option "SHMConfig" "on"
> Option "Vendor" "Sysp"
> Option "ZAxisMapping" "4 5"
> EndSection
>
> Section "ServerLayout"
> Identifier "Layout[all]"
> InputDevice "Keyboard[0]" "CoreKeyboard"
> InputDevice "Mouse[1]" "CorePointer"
> InputDevice "Mouse[3]" "SendCoreEvents"
> Option "Clone" "off"
> Option "Xinerama" "off"
> Screen "Screen[0]"
> EndSection
>

Ok, so you are using Synaptics driver. Does you have evdev module loaded
and do you have /dev/input/eventX nodes created? If not that would explain
why the touchpad is not working in X.

--
Dmitry

2005-11-26 03:54:59

[permalink] [raw]

Subject: Re: psmouse unusable in -mm series (was: 2.6.15-rc1-mm2 unsusable on DELL Inspiron 8200, 2.6.15-rc1 works fine)

On Thursday 24 November 2005 16:20, Rafael J. Wysocki wrote:
> Update:
>
> On Thursday, 24 of November 2005 21:23, Rafael J. Wysocki wrote:
> > On Thursday, 24 of November 2005 13:44, Marc Koschewski wrote:
> > }-- snip --{
> > > > It looks like you are seeing a different bug. The one opened for debian user space
> > > > covers mousedev not being loaded if the kernel is 2.6.15, which leads to no /dev/input
> > > >
> > >
> > > That's what I think, thus the report on LKLM. But noone but me seems to
> > > be trapped into it until... :/
> >
> > FWIW, my touchpad doesn't work with -rc2-mm1 too (usually I use a USB mouse,
> > so I didn't notice before). Here's what dmesg says about it:
> >
> > Synaptics Touchpad, model: 1, fw: 5.9, id: 0x926eb1, caps: 0x804719/0x0
> > input: SynPS/2 Synaptics TouchPad as /class/input/input2
> >
> > The box is an Asus L5D (x86-64).
>
> Actually, it works on the console (ie with gpm), but X is unable to use it,
> apparently. However it used to be, at least on 2.6.14-git9 (this is the latest
> non-mm kernel I've been able to test quickly on this box).
>

Rafael,

does reverting the following patch makes touchpad work?

--
Dmitry

Input: evdev - consolidate compat and regular code

Compat and normal code mirror each other and are hard to maintain.
When EV_SW was added compat_ioctl case was missed. Here is my attempt
at consolidating the code.

Signed-off-by: Dmitry Torokhov <[email protected]>
---

drivers/input/evdev.c | 491 +++++++++++++++++++++-----------------------------
1 files changed, 212 insertions(+), 279 deletions(-)

Index: work/drivers/input/evdev.c
===================================================================
--- work.orig/drivers/input/evdev.c
+++ work/drivers/input/evdev.c
@@ -146,6 +146,7 @@ static int evdev_open(struct inode * ino
}

#ifdef CONFIG_COMPAT
+
struct input_event_compat {
struct compat_timeval time;
__u16 type;
@@ -165,98 +166,107 @@ struct input_event_compat {
# define COMPAT_TEST test_thread_flag(TIF_32BIT)
#endif

-static ssize_t evdev_write_compat(struct file * file, const char __user * buffer, size_t count, loff_t *ppos)
+static inline size_t evdev_event_size(void)
{
- struct evdev_list *list = file->private_data;
- struct input_event_compat event;
- int retval = 0;
+ return COMPAT_TEST ?
+ sizeof(struct input_event_compat) : sizeof(struct input_event);
+}

- while (retval < count) {
- if (copy_from_user(&event, buffer + retval, sizeof(struct input_event_compat)))
+static int evdev_event_from_user(const char __user *buffer, struct input_event *event)
+{
+ if (COMPAT_TEST) {
+ struct input_event_compat compat_event;
+
+ if (copy_from_user(&compat_event, buffer, sizeof(struct input_event_compat)))
+ return -EFAULT;
+
+ event->time.tv_sec = compat_event.time.tv_sec;
+ event->time.tv_usec = compat_event.time.tv_usec;
+ event->type = compat_event.type;
+ event->code = compat_event.code;
+ event->value = compat_event.value;
+
+ } else {
+ if (copy_from_user(event, buffer, sizeof(struct input_event)))
return -EFAULT;
- input_event(list->evdev->handle.dev, event.type, event.code, event.value);
- retval += sizeof(struct input_event_compat);
}

- return retval;
+ return 0;
}
-#endif

-static ssize_t evdev_write(struct file * file, const char __user * buffer, size_t count, loff_t *ppos)
+static int evdev_event_to_user(const char __user *buffer, struct input_event *event)
{
- struct evdev_list *list = file->private_data;
- struct input_event event;
- int retval = 0;
+ if (COMPAT_TEST) {
+ struct input_event_compat compat_event;

- if (!list->evdev->exist) return -ENODEV;
+ compat_event.time.tv_sec = event->time.tv_sec;
+ compat_event.time.tv_usec = event->time.tv_usec;
+ compat_event.type = event->type;
+ compat_event.code = event->code;
+ compat_event.value = event->value;

-#ifdef CONFIG_COMPAT
- if (COMPAT_TEST)
- return evdev_write_compat(file, buffer, count, ppos);
-#endif
-
- while (retval < count) {
+ if (copy_to_user(&compat_event, buffer, sizeof(struct input_event_compat)))
+ return -EFAULT;

- if (copy_from_user(&event, buffer + retval, sizeof(struct input_event)))
+ } else {
+ if (copy_to_user(event, buffer, sizeof(struct input_event)))
return -EFAULT;
- input_event(list->evdev->handle.dev, event.type, event.code, event.value);
- retval += sizeof(struct input_event);
}

- return retval;
+ return 0;
}

-#ifdef CONFIG_COMPAT
-static ssize_t evdev_read_compat(struct file * file, char __user * buffer, size_t count, loff_t *ppos)
+#else
+
+static inline size_t evdev_event_size(void)
{
- struct evdev_list *list = file->private_data;
- int retval;
+ return sizeof(struct input_event);
+}

- if (count < sizeof(struct input_event_compat))
- return -EINVAL;
+static int evdev_event_from_user(const char __user *buffer, struct input_event *event)
+{
+ if (copy_from_user(event, buffer, sizeof(struct input_event)))
+ return -EFAULT;

- if (list->head == list->tail && list->evdev->exist && (file->f_flags & O_NONBLOCK))
- return -EAGAIN;
+ return 0;
+}

- retval = wait_event_interruptible(list->evdev->wait,
- list->head != list->tail || (!list->evdev->exist));
+static int evdev_event_to_user(char __user *buffer, struct input_event *event)
+{
+ if (copy_to_user(buffer, event, sizeof(struct input_event)))
+ return -EFAULT;

- if (retval)
- return retval;
+ return 0;
+}
+
+#endif /* CONFIG_COMPAT */
+
+static ssize_t evdev_write(struct file * file, const char __user * buffer, size_t count, loff_t *ppos)
+{
+ struct evdev_list *list = file->private_data;
+ struct input_event event;
+ int retval = 0;

if (!list->evdev->exist)
return -ENODEV;

- while (list->head != list->tail && retval + sizeof(struct input_event_compat) <= count) {
- struct input_event *event = (struct input_event *) list->buffer + list->tail;
- struct input_event_compat event_compat;
- event_compat.time.tv_sec = event->time.tv_sec;
- event_compat.time.tv_usec = event->time.tv_usec;
- event_compat.type = event->type;
- event_compat.code = event->code;
- event_compat.value = event->value;
+ while (retval < count) {

- if (copy_to_user(buffer + retval, &event_compat,
- sizeof(struct input_event_compat))) return -EFAULT;
- list->tail = (list->tail + 1) & (EVDEV_BUFFER_SIZE - 1);
- retval += sizeof(struct input_event_compat);
+ if (evdev_event_from_user(buffer + retval, &event))
+ return -EFAULT;
+ input_event(list->evdev->handle.dev, event.type, event.code, event.value);
+ retval += evdev_event_size();
}

return retval;
}
-#endif

static ssize_t evdev_read(struct file * file, char __user * buffer, size_t count, loff_t *ppos)
{
struct evdev_list *list = file->private_data;
int retval;

-#ifdef CONFIG_COMPAT
- if (COMPAT_TEST)
- return evdev_read_compat(file, buffer, count, ppos);
-#endif
-
- if (count < sizeof(struct input_event))
+ if (count < evdev_event_size())
return -EINVAL;

if (list->head == list->tail && list->evdev->exist && (file->f_flags & O_NONBLOCK))
@@ -271,11 +281,15 @@ static ssize_t evdev_read(struct file *
if (!list->evdev->exist)
return -ENODEV;

- while (list->head != list->tail && retval + sizeof(struct input_event) <= count) {
- if (copy_to_user(buffer + retval, list->buffer + list->tail,
- sizeof(struct input_event))) return -EFAULT;
+ while (list->head != list->tail && retval + evdev_event_size() <= count) {
+
+ struct input_event *event = (struct input_event *) list->buffer + list->tail;
+
+ if (evdev_event_to_user(buffer + retval, event))
+ return -EFAULT;
+
list->tail = (list->tail + 1) & (EVDEV_BUFFER_SIZE - 1);
- retval += sizeof(struct input_event);
+ retval += evdev_event_size();
}

return retval;
@@ -290,17 +304,95 @@ static unsigned int evdev_poll(struct fi
(list->evdev->exist ? 0 : (POLLHUP | POLLERR));
}

-static long evdev_ioctl(struct file *file, unsigned int cmd, unsigned long arg)
+#ifdef CONFIG_COMPAT
+
+#define BITS_PER_LONG_COMPAT (sizeof(compat_long_t) * 8)
+#define NBITS_COMPAT(x) ((((x) - 1) / BITS_PER_LONG_COMPAT) + 1)
+
+#ifdef __BIG_ENDIAN
+static int bits_to_user(unsigned long *bits, unsigned int maxbit,
+ unsigned int maxlen, void __user *p, int compat)
+{
+ int len, i;
+
+ if (compat) {
+ len = NBITS_COMPAT(maxbit) * sizeof(compat_long_t);
+ if (len < maxlen)
+ len = maxlen;
+
+ for (i = 0; i < len / sizeof(compat_long_t); i++)
+ if (copy_to_user((compat_long_t __user *) p + i,
+ (compat_long_t *) bits +
+ i + 1 - ((i % 2) << 1),
+ sizeof(compat_long_t)))
+ return -EFAULT;
+ } else {
+ len = NBITS(maxbit) * sizeof(long);
+ if (len > maxlen)
+ len = maxlen;
+
+ if (copy_to_user(p, bits, len))
+ return -EFAULT;
+ }
+
+ return len;
+}
+#else
+static int bits_to_user(unsigned long *bits, unsigned int maxbit,
+ unsigned int maxlen, void __user *p, int compat)
+{
+ int len = compat ?
+ NBITS_COMPAT(maxbit) * sizeof(compat_long_t) :
+ NBITS(maxbit) * sizeof(long);
+
+ if (len > maxlen)
+ len = maxlen;
+
+ return copy_to_user(p, bits, len) ? -EFAULT : len;
+}
+#endif /* __BIG_ENDIAN */
+
+#else
+
+static int bits_to_user(unsigned long *bits, unsigned int maxbit,
+ unsigned int maxlen, void __user *p, int compat)
+{
+ int len = NBITS(maxbit) * sizeof(long);
+
+ if (len > maxlen)
+ len = maxlen;
+
+ return copy_to_user(p, bits, len) ? -EFAULT : len;
+}
+
+#endif /* CONFIG_COMPAT */
+
+static int str_to_user(const char *str, unsigned int maxlen, void __user *p)
+{
+ int len;
+
+ if (!str)
+ return -ENOENT;
+
+ len = strlen(str) + 1;
+ if (len > maxlen)
+ len = maxlen;
+
+ return copy_to_user(p, str, len) ? -EFAULT : len;
+}
+
+static long evdev_ioctl_handler(struct file *file, unsigned int cmd,
+ void __user *p, int compat_mode)
{
struct evdev_list *list = file->private_data;
struct evdev *evdev = list->evdev;
struct input_dev *dev = evdev->handle.dev;
struct input_absinfo abs;
- void __user *p = (void __user *)arg;
- int __user *ip = (int __user *)arg;
+ int __user *ip = (int __user *)p;
int i, t, u, v;

- if (!evdev->exist) return -ENODEV;
+ if (!evdev->exist)
+ return -ENODEV;

switch (cmd) {

@@ -308,26 +400,39 @@ static long evdev_ioctl(struct file *fil
return put_user(EV_VERSION, ip);

case EVIOCGID:
- return copy_to_user(p, &dev->id, sizeof(struct input_id)) ? -EFAULT : 0;
+ if (copy_to_user(p, &dev->id, sizeof(struct input_id)))
+ return -EFAULT;
+
+ return 0;

case EVIOCGKEYCODE:
- if (get_user(t, ip)) return -EFAULT;
- if (t < 0 || t >= dev->keycodemax || !dev->keycodesize) return -EINVAL;
- if (put_user(INPUT_KEYCODE(dev, t), ip + 1)) return -EFAULT;
+ if (get_user(t, ip))
+ return -EFAULT;
+ if (t < 0 || t >= dev->keycodemax || !dev->keycodesize)
+ return -EINVAL;
+ if (put_user(INPUT_KEYCODE(dev, t), ip + 1))
+ return -EFAULT;
return 0;

case EVIOCSKEYCODE:
- if (get_user(t, ip)) return -EFAULT;
- if (t < 0 || t >= dev->keycodemax || !dev->keycodesize) return -EINVAL;
- if (get_user(v, ip + 1)) return -EFAULT;
- if (v < 0 || v > KEY_MAX) return -EINVAL;
- if (dev->keycodesize < sizeof(v) && (v >> (dev->keycodesize * 8))) return -EINVAL;
+ if (get_user(t, ip))
+ return -EFAULT;
+ if (t < 0 || t >= dev->keycodemax || !dev->keycodesize)
+ return -EINVAL;
+ if (get_user(v, ip + 1))
+ return -EFAULT;
+ if (v < 0 || v > KEY_MAX)
+ return -EINVAL;
+ if (dev->keycodesize < sizeof(v) && (v >> (dev->keycodesize * 8)))
+ return -EINVAL;
+
u = SET_INPUT_KEYCODE(dev, t, v);
clear_bit(u, dev->keybit);
set_bit(v, dev->keybit);
for (i = 0; i < dev->keycodemax; i++)
- if (INPUT_KEYCODE(dev,i) == u)
+ if (INPUT_KEYCODE(dev, i) == u)
set_bit(u, dev->keybit);
+
return 0;

case EVIOCSFF:
@@ -338,17 +443,17 @@ static long evdev_ioctl(struct file *fil
if (copy_from_user(&effect, p, sizeof(effect)))
return -EFAULT;
err = dev->upload_effect(dev, &effect);
- if (put_user(effect.id, &(((struct ff_effect __user *)arg)->id)))
+ if (put_user(effect.id, &(((struct ff_effect __user *)p)->id)))
return -EFAULT;
return err;
- }
- else return -ENOSYS;
+ } else
+ return -ENOSYS;

case EVIOCRMFF:
- if (dev->erase_effect) {
- return dev->erase_effect(dev, (int)arg);
- }
- else return -ENOSYS;
+ if (!dev->erase_effect)
+ return -ENOSYS;
+
+ return dev->erase_effect(dev, (int)(unsigned long) p);

case EVIOCGEFFECTS:
if (put_user(dev->ff_effects_max, ip))
@@ -356,7 +461,7 @@ static long evdev_ioctl(struct file *fil
return 0;

case EVIOCGRAB:
- if (arg) {
+ if (p) {
if (evdev->grab)
return -EBUSY;
if (input_grab_device(&evdev->handle))
@@ -395,62 +500,33 @@ static long evdev_ioctl(struct file *fil
case EV_SW: bits = dev->swbit; len = SW_MAX; break;
default: return -EINVAL;
}
- len = NBITS(len) * sizeof(long);
- if (len > _IOC_SIZE(cmd)) len = _IOC_SIZE(cmd);
- return copy_to_user(p, bits, len) ? -EFAULT : len;
+ return bits_to_user(bits, len, _IOC_SIZE(cmd), p, compat_mode);
}

- if (_IOC_NR(cmd) == _IOC_NR(EVIOCGKEY(0))) {
- int len;
- len = NBITS(KEY_MAX) * sizeof(long);
- if (len > _IOC_SIZE(cmd)) len = _IOC_SIZE(cmd);
- return copy_to_user(p, dev->key, len) ? -EFAULT : len;
- }
+ if (_IOC_NR(cmd) == _IOC_NR(EVIOCGKEY(0)))
+ return bits_to_user(dev->key, KEY_MAX, _IOC_SIZE(cmd),
+ p, compat_mode);

- if (_IOC_NR(cmd) == _IOC_NR(EVIOCGLED(0))) {
- int len;
- len = NBITS(LED_MAX) * sizeof(long);
- if (len > _IOC_SIZE(cmd)) len = _IOC_SIZE(cmd);
- return copy_to_user(p, dev->led, len) ? -EFAULT : len;
- }
+ if (_IOC_NR(cmd) == _IOC_NR(EVIOCGLED(0)))
+ return bits_to_user(dev->led, LED_MAX, _IOC_SIZE(cmd),
+ p, compat_mode);

- if (_IOC_NR(cmd) == _IOC_NR(EVIOCGSND(0))) {
- int len;
- len = NBITS(SND_MAX) * sizeof(long);
- if (len > _IOC_SIZE(cmd)) len = _IOC_SIZE(cmd);
- return copy_to_user(p, dev->snd, len) ? -EFAULT : len;
- }
+ if (_IOC_NR(cmd) == _IOC_NR(EVIOCGSND(0)))
+ return bits_to_user(dev->snd, SND_MAX, _IOC_SIZE(cmd),
+ p, compat_mode);

- if (_IOC_NR(cmd) == _IOC_NR(EVIOCGSW(0))) {
- int len;
- len = NBITS(SW_MAX) * sizeof(long);
- if (len > _IOC_SIZE(cmd)) len = _IOC_SIZE(cmd);
- return copy_to_user(p, dev->sw, len) ? -EFAULT : len;
- }
+ if (_IOC_NR(cmd) == _IOC_NR(EVIOCGSW(0)))
+ return bits_to_user(dev->sw, SW_MAX, _IOC_SIZE(cmd),
+ p, compat_mode);

- if (_IOC_NR(cmd) == _IOC_NR(EVIOCGNAME(0))) {
- int len;
- if (!dev->name) return -ENOENT;
- len = strlen(dev->name) + 1;
- if (len > _IOC_SIZE(cmd)) len = _IOC_SIZE(cmd);
- return copy_to_user(p, dev->name, len) ? -EFAULT : len;
- }
+ if (_IOC_NR(cmd) == _IOC_NR(EVIOCGNAME(0)))
+ return str_to_user(dev->name, _IOC_SIZE(cmd), p);

- if (_IOC_NR(cmd) == _IOC_NR(EVIOCGPHYS(0))) {
- int len;
- if (!dev->phys) return -ENOENT;
- len = strlen(dev->phys) + 1;
- if (len > _IOC_SIZE(cmd)) len = _IOC_SIZE(cmd);
- return copy_to_user(p, dev->phys, len) ? -EFAULT : len;
- }
+ if (_IOC_NR(cmd) == _IOC_NR(EVIOCGPHYS(0)))
+ return str_to_user(dev->phys, _IOC_SIZE(cmd), p);

- if (_IOC_NR(cmd) == _IOC_NR(EVIOCGUNIQ(0))) {
- int len;
- if (!dev->uniq) return -ENOENT;
- len = strlen(dev->uniq) + 1;
- if (len > _IOC_SIZE(cmd)) len = _IOC_SIZE(cmd);
- return copy_to_user(p, dev->uniq, len) ? -EFAULT : len;
- }
+ if (_IOC_NR(cmd) == _IOC_NR(EVIOCGUNIQ(0)))
+ return str_to_user(dev->uniq, _IOC_SIZE(cmd), p);

if ((_IOC_NR(cmd) & ~ABS_MAX) == _IOC_NR(EVIOCGABS(0))) {

@@ -492,158 +568,15 @@ static long evdev_ioctl(struct file *fil
return -EINVAL;
}

-#ifdef CONFIG_COMPAT
-
-#define BITS_PER_LONG_COMPAT (sizeof(compat_long_t) * 8)
-#define NBITS_COMPAT(x) ((((x)-1)/BITS_PER_LONG_COMPAT)+1)
-#define OFF_COMPAT(x) ((x)%BITS_PER_LONG_COMPAT)
-#define BIT_COMPAT(x) (1UL<<OFF_COMPAT(x))
-#define LONG_COMPAT(x) ((x)/BITS_PER_LONG_COMPAT)
-#define test_bit_compat(bit, array) ((array[LONG_COMPAT(bit)] >> OFF_COMPAT(bit)) & 1)
-
-#ifdef __BIG_ENDIAN
-#define bit_to_user(bit, max) \
-do { \
- int i; \
- int len = NBITS_COMPAT((max)) * sizeof(compat_long_t); \
- if (len > _IOC_SIZE(cmd)) len = _IOC_SIZE(cmd); \
- for (i = 0; i < len / sizeof(compat_long_t); i++) \
- if (copy_to_user((compat_long_t __user *) p + i, \
- (compat_long_t*) (bit) + i + 1 - ((i % 2) << 1), \
- sizeof(compat_long_t))) \
- return -EFAULT; \
- return len; \
-} while (0)
-#else
-#define bit_to_user(bit, max) \
-do { \
- int len = NBITS_COMPAT((max)) * sizeof(compat_long_t); \
- if (len > _IOC_SIZE(cmd)) len = _IOC_SIZE(cmd); \
- return copy_to_user(p, (bit), len) ? -EFAULT : len; \
-} while (0)
-#endif
+static long evdev_ioctl(struct file *file, unsigned int cmd, unsigned long arg)
+{
+ return evdev_ioctl_handler(file, cmd, (void __user *)arg, 0);
+}

+#ifdef CONFIG_COMPAT
static long evdev_ioctl_compat(struct file *file, unsigned int cmd, unsigned long arg)
{
- struct evdev_list *list = file->private_data;
- struct evdev *evdev = list->evdev;
- struct input_dev *dev = evdev->handle.dev;
- struct input_absinfo abs;
- void __user *p = compat_ptr(arg);
-
- if (!evdev->exist) return -ENODEV;
-
- switch (cmd) {
-
- case EVIOCGVERSION:
- case EVIOCGID:
- case EVIOCGKEYCODE:
- case EVIOCSKEYCODE:
- case EVIOCSFF:
- case EVIOCRMFF:
- case EVIOCGEFFECTS:
- case EVIOCGRAB:
- return evdev_ioctl(file, cmd, (unsigned long) p);
-
- default:
-
- if (_IOC_TYPE(cmd) != 'E')
- return -EINVAL;
-
- if (_IOC_DIR(cmd) == _IOC_READ) {
-
- if ((_IOC_NR(cmd) & ~EV_MAX) == _IOC_NR(EVIOCGBIT(0,0))) {
- long *bits;
- int max;
-
- switch (_IOC_NR(cmd) & EV_MAX) {
- case 0: bits = dev->evbit; max = EV_MAX; break;
- case EV_KEY: bits = dev->keybit; max = KEY_MAX; break;
- case EV_REL: bits = dev->relbit; max = REL_MAX; break;
- case EV_ABS: bits = dev->absbit; max = ABS_MAX; break;
- case EV_MSC: bits = dev->mscbit; max = MSC_MAX; break;
- case EV_LED: bits = dev->ledbit; max = LED_MAX; break;
- case EV_SND: bits = dev->sndbit; max = SND_MAX; break;
- case EV_FF: bits = dev->ffbit; max = FF_MAX; break;
- case EV_SW: bits = dev->swbit; max = SW_MAX; break;
- default: return -EINVAL;
- }
- bit_to_user(bits, max);
- }
-
- if (_IOC_NR(cmd) == _IOC_NR(EVIOCGKEY(0)))
- bit_to_user(dev->key, KEY_MAX);
-
- if (_IOC_NR(cmd) == _IOC_NR(EVIOCGLED(0)))
- bit_to_user(dev->led, LED_MAX);
-
- if (_IOC_NR(cmd) == _IOC_NR(EVIOCGSND(0)))
- bit_to_user(dev->snd, SND_MAX);
-
- if (_IOC_NR(cmd) == _IOC_NR(EVIOCGSW(0)))
- bit_to_user(dev->sw, SW_MAX);
-
- if (_IOC_NR(cmd) == _IOC_NR(EVIOCGNAME(0))) {
- int len;
- if (!dev->name) return -ENOENT;
- len = strlen(dev->name) + 1;
- if (len > _IOC_SIZE(cmd)) len = _IOC_SIZE(cmd);
- return copy_to_user(p, dev->name, len) ? -EFAULT : len;
- }
-
- if (_IOC_NR(cmd) == _IOC_NR(EVIOCGPHYS(0))) {
- int len;
- if (!dev->phys) return -ENOENT;
- len = strlen(dev->phys) + 1;
- if (len > _IOC_SIZE(cmd)) len = _IOC_SIZE(cmd);
- return copy_to_user(p, dev->phys, len) ? -EFAULT : len;
- }
-
- if (_IOC_NR(cmd) == _IOC_NR(EVIOCGUNIQ(0))) {
- int len;
- if (!dev->uniq) return -ENOENT;
- len = strlen(dev->uniq) + 1;
- if (len > _IOC_SIZE(cmd)) len = _IOC_SIZE(cmd);
- return copy_to_user(p, dev->uniq, len) ? -EFAULT : len;
- }
-
- if ((_IOC_NR(cmd) & ~ABS_MAX) == _IOC_NR(EVIOCGABS(0))) {
-
- int t = _IOC_NR(cmd) & ABS_MAX;
-
- abs.value = dev->abs[t];
- abs.minimum = dev->absmin[t];
- abs.maximum = dev->absmax[t];
- abs.fuzz = dev->absfuzz[t];
- abs.flat = dev->absflat[t];
-
- if (copy_to_user(p, &abs, sizeof(struct input_absinfo)))
- return -EFAULT;
-
- return 0;
- }
- }
-
- if (_IOC_DIR(cmd) == _IOC_WRITE) {
-
- if ((_IOC_NR(cmd) & ~ABS_MAX) == _IOC_NR(EVIOCSABS(0))) {
-
- int t = _IOC_NR(cmd) & ABS_MAX;
-
- if (copy_from_user(&abs, p, sizeof(struct input_absinfo)))
- return -EFAULT;
-
- dev->abs[t] = abs.value;
- dev->absmin[t] = abs.minimum;
- dev->absmax[t] = abs.maximum;
- dev->absfuzz[t] = abs.fuzz;
- dev->absflat[t] = abs.flat;
-
- return 0;
- }
- }
- }
- return -EINVAL;
+ return evdev_ioctl_handler(file, cmd, compat_ptr(arg), 1);
}
#endif

2005-11-26 04:50:34

[permalink] [raw]

Subject: Re: psmouse unusable in -mm series (was: 2.6.15-rc1-mm2 unsusable on DELL Inspiron 8200, 2.6.15-rc1 works fine)

On Friday 25 November 2005 22:54, Dmitry Torokhov wrote:
> > Actually, it works on the console (ie with gpm), but X is unable to use it,
> > apparently. ?However it used to be, at least on 2.6.14-git9 (this is the latest
> > non-mm kernel I've been able to test quickly on this box).
> >
>
> Rafael,
>
> does reverting the following patch makes touchpad work?

Or, try dropping this patch on top of -mm.

--
Dmitry

Signed-off-by: Dmitry Torokhov <[email protected]>
---

drivers/input/evdev.c | 4 ++--
1 files changed, 2 insertions(+), 2 deletions(-)

Index: work/drivers/input/evdev.c
===================================================================
--- work.orig/drivers/input/evdev.c
+++ work/drivers/input/evdev.c
@@ -205,11 +205,11 @@ static int evdev_event_to_user(const cha
compat_event.code = event->code;
compat_event.value = event->value;

- if (copy_to_user(&compat_event, buffer, sizeof(struct input_event_compat)))
+ if (copy_to_user(buffer, &compat_event, sizeof(struct input_event_compat)))
return -EFAULT;

} else {
- if (copy_to_user(event, buffer, sizeof(struct input_event)))
+ if (copy_to_user(buffer, event, sizeof(struct input_event)))
return -EFAULT;
}

2005-11-26 17:28:00

[permalink] [raw]

Subject: Re: psmouse unusable in -mm series (was: 2.6.15-rc1-mm2 unsusable on DELL Inspiron 8200, 2.6.15-rc1 works fine)

On Saturday, 26 of November 2005 05:50, Dmitry Torokhov wrote:
> On Friday 25 November 2005 22:54, Dmitry Torokhov wrote:
> > > Actually, it works on the console (ie with gpm), but X is unable to use it,
> > > apparently. ?However it used to be, at least on 2.6.14-git9 (this is the latest
> > > non-mm kernel I've been able to test quickly on this box).
> > >
> >
> > Rafael,
> >
> > does reverting the following patch makes touchpad work?
>
> Or, try dropping this patch on top of -mm.

That helps (additionally I've dropped "const" from the header of
evdev_event_to_user, to avoid warnings).

Thanks,
Rafael

2005-11-28 06:01:29