Skip to contentSkip to conversation

2.6.18-rc strange hotplug/udev/uevent problem

DISCUSSION Andrew Clayton 3 participants 6 messages Aug 7, 2006 – Aug 9, 2006

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

Hi,

Got a weird problem here.

On x86 Fedora Core 5 with 2.6.17 with GNOME, plugging in a usb stick would result in it being mounted. With 2.6.18-rc this no longer occurs. FC5 got an update to hal to work with 2.6.18 kernels, but it don't work for me. I'm having the same problem on 3 x86 FC5 machines.

The weird thing is, this all works on my x86-64 FC5 workstation with 2.6.18-rc both before and after the hal update.

Anyway I submitted a bug report against HAL suspecting it broken

https://bugs.freedesktop.org/show_bug.cgi?id=7756

Perhaps not. So I turn my attention more to the kernel.

2.6.17 was working fine. You could plug/unplug/plug a USB memory stick and it would get mounted each time.

2.6.18-rc[23] works the same as above on my x86-64 FC5 box.

2.6.18-rc[23] and 2.6.18-rc3-git7 on x86 built with usb/scsi/sd/vfat/nls_* built as modules will mount on the first plug but not subsequent plugs.

If you rmmod the sd_mod module and plug in, then it will get mounted.

When the kernel is built non-modular then plugging in will not mount it. If you boot with the device plugged in then when you log into GNOME it will be mounted. But subsequent unmount/plug ins will not get it mounted.

The following is from the udevmonitor program running under 2.6.18-rc built modular.

Plugging device in with sd_mod not loaded.

UEVENT[1154789226.207797] add@/devices/pci0000:00/0000:00:11.2/usb1/1-2 UEVENT[1154789226.207985] add@/devices/pci0000:00/0000:00:11.2/usb1/1-2/usbdev1.11_ep00 UEVENT[1154789226.210770] add@/devices/pci0000:00/0000:00:11.2/usb1/1-2/1-2:1.0 UDEV [1154789226.210770] add@/devices/pci0000:00/0000:00:11.2/usb1/1-2 UEVENT[1154789226.211037] add@/class/scsi_host/host9
UEVENT[1154789226.211097] add@/devices/pci0000:00/0000:00:11.2/usb1/1-2/1-2:1.0/usbdev1.11_ep81 UEVENT[1154789226.211157] add@/devices/pci0000:00/0000:00:11.2/usb1/1-2/1-2:1.0/usbdev1.11_ep02 UEVENT[1154789226.211215] add@/devices/pci0000:00/0000:00:11.2/usb1/1-2/usbdev1.11

UDEV  [1154789226.211690] add@/devices/pci0000:00/0000:00:11.2/usb1/1-2/usbdev1.11_ep00
UDEV  [1154789226.452013] add@/devices/pci0000:00/0000:00:11.2/usb1/1-2/usbdev1.11
UDEV  [1154789226.956997] add@/devices/pci0000:00/0000:00:11.2/usb1/1-2/1-2:1.0
UDEV  [1154789226.959419] add@/devices/pci0000:00/0000:00:11.2/usb1/1-2/1-2:1.0/usbdev1.11_ep81
UDEV  [1154789226.959466] add@/devices/pci0000:00/0000:00:11.2/usb1/1-2/1-2:1.0/usbdev1.11_ep02
UDEV  [1154789226.959875] add@/class/scsi_host/host9

UEVENT[1154789231.647515] add@/devices/pci0000:00/0000:00:11.2/usb1/1-2/1-2:1.0/host9/target9:0:0/9:0:0:0 UDEV [1154789231.647515] add@/devices/pci0000:00/0000:00:11.2/usb1/1-2/1-2:1.0/host9/target9:0:0/9:0:0:0 UEVENT[1154789231.647570] add@/class/scsi_device/9:0:0:0
UEVENT[1154789231.749919] add@/module/sd_mod
UEVENT[1154789231.751494] add@/bus/scsi/drivers/sd
UEVENT[1154789231.752839] add@/class/scsi_disk/9:0:0:0
UDEV [1154789231.753751] add@/class/scsi_disk/9:0:0:0
UEVENT[1154789231.832312] add@/block/sda
UEVENT[1154789231.834681] add@/block/sda/sda1

UDEV  [1154789231.836590] add@/class/scsi_device/9:0:0:0
UDEV  [1154789231.933474] add@/block/sda
UDEV  [1154789232.106892] add@/block/sda/sda1

UEVENT[1154789235.937510] mount@/block/sda/sda1
UDEV [1154789235.937510] mount@/block/sda/sda1

Plugging device in with sd_mod loaded.

UEVENT[1154789359.096963] add@/devices/pci0000:00/0000:00:11.2/usb1/1-2
UDEV [1154789359.096963] add@/devices/pci0000:00/0000:00:11.2/usb1/1-2 UEVENT[1154789359.097052] add@/devices/pci0000:00/0000:00:11.2/usb1/1-2/usbdev1.12_ep00 UDEV [1154789359.098126] add@/devices/pci0000:00/0000:00:11.2/usb1/1-2/usbdev1.12_ep00 UEVENT[1154789359.100369] add@/devices/pci0000:00/0000:00:11.2/usb1/1-2/1-2:1.0 UEVENT[1154789359.100418] add@/class/scsi_host/host10
UEVENT[1154789359.100427] add@/devices/pci0000:00/0000:00:11.2/usb1/1-2/1-2:1.0/usbdev1.12_ep81 UEVENT[1154789359.100435] add@/devices/pci0000:00/0000:00:11.2/usb1/1-2/1-2:1.0/usbdev1.12_ep02 UEVENT[1154789359.100444] add@/devices/pci0000:00/0000:00:11.2/usb1/1-2/usbdev1.12

UDEV  [1154789359.106148] add@/devices/pci0000:00/0000:00:11.2/usb1/1-2/usbdev1.12
UDEV  [1154789359.187500] add@/devices/pci0000:00/0000:00:11.2/usb1/1-2/1-2:1.0
UDEV  [1154789359.189198] add@/class/scsi_host/host10
UDEV  [1154789359.190078] add@/devices/pci0000:00/0000:00:11.2/usb1/1-2/1-2:1.0/usbdev1.12_ep81
UDEV  [1154789359.191020] add@/devices/pci0000:00/0000:00:11.2/usb1/1-2/1-2:1.0/usbdev1.12_ep02

UEVENT[1154789364.540249] add@/devices/pci0000:00/0000:00:11.2/usb1/1-2/1-2:1.0/host10/target10:0:0/10:0:0:0 UDEV [1154789364.540249] add@/devices/pci0000:00/0000:00:11.2/usb1/1-2/1-2:1.0/host10/target10:0:0/10:0:0:0 UEVENT[1154789364.540355] add@/class/scsi_disk/10:0:0:0
UDEV [1154789364.541272] add@/class/scsi_disk/10:0:0:0
UEVENT[1154789364.586896] add@/block/sda
UEVENT[1154789364.586954] add@/block/sda/sda1
UEVENT[1154789364.586963] add@/class/scsi_device/10:0:0:0

UDEV  [1154789364.598962] add@/class/scsi_device/10:0:0:0
UDEV  [1154789364.719350] add@/block/sda
UDEV  [1154789364.874149] add@/block/sda/sda1

No mount events here, this is similar output to the non-modular case..

Here's udevmonitor output from my x86-64 FC5 2.6.18-rc[23]

UEVENT[1154510664.116702] add@/devices/pci0000:00/0000:00:10.4/usb1/1-8 UEVENT[1154510664.116905] add@/devices/pci0000:00/0000:00:10.4/usb1/1-8/usbdev1.12_ep00 UEVENT[1154510664.116911] add@/devices/pci0000:00/0000:00:10.4/usb1/1-8/1-8:1.0 UEVENT[1154510664.116915] add@/class/scsi_host/host12
UEVENT[1154510664.116919] add@/devices/pci0000:00/0000:00:10.4/usb1/1-8/1-8:1.0/usbdev1.12_ep81 UEVENT[1154510664.116924] add@/devices/pci0000:00/0000:00:10.4/usb1/1-8/1-8:1.0/usbdev1.12_ep02 UEVENT[1154510664.116928] add@/devices/pci0000:00/0000:00:10.4/usb1/1-8/1-8:1.0/usbdev1.12_ep83 UEVENT[1154510664.116932] add@/devices/pci0000:00/0000:00:10.4/usb1/1-8/usbdev1.12

UDEV  [1154510664.200847] add@/devices/pci0000:00/0000:00:10.4/usb1/1-8
UDEV  [1154510664.364513] add@/devices/pci0000:00/0000:00:10.4/usb1/1-8/usbdev1.12_ep00
UDEV  [1154510664.371420] add@/devices/pci0000:00/0000:00:10.4/usb1/1-8/usbdev1.12
UDEV  [1154510664.631746] add@/devices/pci0000:00/0000:00:10.4/usb1/1-8/1-8:1.0
UDEV  [1154510664.652152] add@/class/scsi_host/host12
UDEV  [1154510664.776377] add@/devices/pci0000:00/0000:00:10.4/usb1/1-8/1-8:1.0/usbdev1.12_ep81
UDEV  [1154510664.785827] add@/devices/pci0000:00/0000:00:10.4/usb1/1-8/1-8:1.0/usbdev1.12_ep02
UDEV  [1154510664.795277] add@/devices/pci0000:00/0000:00:10.4/usb1/1-8/1-8:1.0/usbdev1.12_ep83

UEVENT[1154510669.116889] add@/devices/pci0000:00/0000:00:10.4/usb1/1-8/1-8:1.0/host12/target12:0:0/12:0:0:0 UEVENT[1154510669.116915] add@/class/scsi_disk/12:0:0:0
UDEV [1154510669.118308] add@/devices/pci0000:00/0000:00:10.4/usb1/1-8/1-8:1.0/host12/target12:0:0/12:0:0:0 UDEV [1154510669.136236] add@/class/scsi_disk/12:0:0:0
UEVENT[1154510670.246094] add@/block/sdb
UEVENT[1154510670.246115] add@/block/sdb/sdb1
UEVENT[1154510670.246119] add@/class/scsi_device/12:0:0:0

UDEV  [1154510670.250690] add@/class/scsi_device/12:0:0:0
UDEV  [1154510670.383650] add@/block/sdb
UDEV  [1154510670.555570] add@/block/sdb/sdb1

UEVENT[1154510671.433544] mount@/block/sdb/sdb1
UDEV [1154510671.434129] mount@/block/sdb/sdb1

I guess the next thing is to try and narrow down when this started, but as that could take some time, I'm posting this now.

Cheers,

Andrew

#1 lore raw
Quoted 22 lines · Andrew Clayton

Hi,

Got a weird problem here.

On x86 Fedora Core 5 with 2.6.17 with GNOME, plugging in a usb stick would result in it being mounted. With 2.6.18-rc this no longer occurs. FC5 got an update to hal to work with 2.6.18 kernels, but it don't work for me. I'm having the same problem on 3 x86 FC5 machines.

The weird thing is, this all works on my x86-64 FC5 workstation with 2.6.18-rc both before and after the hal update.

Anyway I submitted a bug report against HAL suspecting it broken

https://bugs.freedesktop.org/show_bug.cgi?id=7756

Perhaps not. So I turn my attention more to the kernel.

2.6.17 was working fine. You could plug/unplug/plug a USB memory stick and it would get mounted each time.

Quoted 7 lines

2.6.18-rc[23] works the same as above on my x86-64 FC5 box.

2.6.18-rc[23] and 2.6.18-rc3-git7 on x86 built with usb/scsi/sd/vfat/nls_* built as modules will mount on the first plug but not subsequent plugs.

If you rmmod the sd_mod module and plug in, then it will get mounted.

That's just wierd. I can't think of anything that has changed recently to cause this.

Can you use 'git bisect' to try to narrow it down which change caused the problem?

thansk,

greg k-h

#2 lore raw
Quoted 7 lines · Greg KH
Quoted 4 lines · Andrew Clayton

plug but not subsequent plugs.

If you rmmod the sd_mod module and plug in, then it will get mounted.

That's just wierd. I can't think of anything that has changed

Heh, yeah I thought so.

Quoted 4 lines

recently to cause this.

Can you use 'git bisect' to try to narrow it down which change caused the problem?

Yeah, I'll start looking tonight.

Quoted 3 lines

thansk,

greg k-h

Cheers,

Andrew

#3 lore raw
In reply to Greg KH
Quoted 10 lines · Greg KH
Quoted 2 lines · Andrew Clayton

If you rmmod the sd_mod module and plug in, then it will get mounted.

That's just wierd. I can't think of anything that has changed recently to cause this.

Can you use 'git bisect' to try to narrow it down which change caused the problem?

I've done a binary search of kernels and found that the breakage occurs between 2.6.17-git3 and 2.6.17-git4

Next I'll figure out git bisect to try and narrow it further.

Quoted 3 lines

thansk,

greg k-h

Cheers,

Andrew

#4 lore raw
In reply to Greg KH
Quoted 2 lines · Greg KH

Can you use 'git bisect' to try to narrow it down which change caused the problem?

I did

git bisect start drivers/scsi drivers/usb
git bad v2.6.18-rc1
git good v2.6.17

Heres what git bisect came up with

c32ba30f76eb18b3d4449072fe9c345a9574796b is first bad commit commit c32ba30f76eb18b3d4449072fe9c345a9574796b

Author: Paul Serice <[email protected]>
Date:   Wed Jun 7 10:23:38 2006 -0700

[PATCH] USB: EHCI works again on NVidia controllers with >2GB RAM

From: Paul Serice <[email protected]>

The workaround in commit f7201c3dcd7799f2aa3d6ec427b194225360ecee broke. The work around requires memory for DMA transfers for some NVidia EHCI controllers to be below 2GB, but recent changes have caused some DMA memory to be allocated before the DMA mask is set.

:040000 040000 754a9f8cccedc71e8e3689f2b0dee811d94fdc11 5559c0eafe042377430052272d027f0458805126 M drivers

What would be the way to revert that patch from 2.6.18-rc4 to confirm it's the culprit?

Heres the output from git bisect log

git-bisect start 'drivers/scsi' 'drivers/usb'
# bad: [7df8ea909888d4856d3aff1c41192739d715a393] Linux 2.6.18-rc1
git-bisect bad 7df8ea909888d4856d3aff1c41192739d715a393
# good: [8ba130df4b67fa40878ccf80d54615132d24bc68] Linux v2.6.17
git-bisect good 8ba130df4b67fa40878ccf80d54615132d24bc68
# good: [d7a80dad2fe19a2b8c119c8e9cba605474a75a2b] libata: convert several bmdma-style controllers to new EH, take #3 git-bisect good d7a80dad2fe19a2b8c119c8e9cba605474a75a2b
# bad: [d5681fe8110e1169902df37a8fd8bd01c2b7810e] USB: unusual_devs entry for Nokia N80
git-bisect bad d5681fe8110e1169902df37a8fd8bd01c2b7810e
# bad: [7327413c745c2f8e8d4b92f76759821263b095c1] USB: added support for ASIX 88178 chipset USB Gigabit Ethernet adaptor git-bisect bad 7327413c745c2f8e8d4b92f76759821263b095c1
# good: [cb63067a772c0149184309a1f232d62c81a93673] [SCSI] qla2xxx: Consolidate "qla2xxx" string usage to a #define. git-bisect good cb63067a772c0149184309a1f232d62c81a93673
# good: [beb40487508290f5d6565598c60a3f44261beef2] [SCSI] remove scsi_request infrastructure git-bisect good beb40487508290f5d6565598c60a3f44261beef2
# good: [67d59dfdeb21df2c16dcd478b66177e91178ecd0] [SCSI] 53c700: remove reliance on deprecated cmnd fields git-bisect good 67d59dfdeb21df2c16dcd478b66177e91178ecd0
# good: [c6387a48cf5958e43c201fc27a158c328927531a] [SPARC]: Kill __irq_itoa().
git-bisect good c6387a48cf5958e43c201fc27a158c328927531a
# bad: [5cd330f4f30eb154e7af9210bed42dfcd447997f] USB: add YEALINK phones to the HID_QUIRK_IGNORE blacklist git-bisect bad 5cd330f4f30eb154e7af9210bed42dfcd447997f
# bad: [c32ba30f76eb18b3d4449072fe9c345a9574796b] USB: EHCI works again on NVidia controllers with >2GB RAM git-bisect bad c32ba30f76eb18b3d4449072fe9c345a9574796b
# good: [28e4b224955cbe30275b2a7842e729023a4f4b03] Merge master.kernel.org:/pub/scm/linux/kernel/git/jejb/scsi-misc-2.6 git-bisect good 28e4b224955cbe30275b2a7842e729023a4f4b03

Quoted 3 lines

thansk,

greg k-h

Cheers,

Andrew

#5 lore raw
Quoted 27 lines · On Wed, 9 Aug 2006 01:41:04 +0100 Andrew Clayton <[email protected]>…
Quoted 2 lines · Greg KH

Can you use 'git bisect' to try to narrow it down which change caused the problem?

I did

git bisect start drivers/scsi drivers/usb
git bad v2.6.18-rc1
git good v2.6.17

Heres what git bisect came up with

c32ba30f76eb18b3d4449072fe9c345a9574796b is first bad commit commit c32ba30f76eb18b3d4449072fe9c345a9574796b

Author: Paul Serice <[email protected]>
Date:   Wed Jun 7 10:23:38 2006 -0700

[PATCH] USB: EHCI works again on NVidia controllers with >2GB RAM

From: Paul Serice <[email protected]>

The workaround in commit f7201c3dcd7799f2aa3d6ec427b194225360ecee broke. The work around requires memory for DMA transfers for some NVidia EHCI controllers to be below 2GB, but recent changes have caused some DMA memory to be allocated before the DMA mask is set.

Quoted 9 lines

:040000 040000 754a9f8cccedc71e8e3689f2b0dee811d94fdc11 5559c0eafe042377430052272d027f0458805126 M drivers

What would be the way to revert that patch from 2.6.18-rc4 to confirm it's the culprit?

Here's a backout patch, against 2.6.18-rc4:

From: Andrew Morton <[email protected]>

Revert c32ba30f76eb18b3d4449072fe9c345a9574796b: it broke Andrew Clayton's machine.

Cc Andrew Clayton <[email protected]>
Cc Paul Serice <[email protected]>
Cc David Brownell <[email protected]>
Cc Greg Kroah-Hartman <[email protected]>

---

 drivers/usb/host/ehci-pci.c |   39 +++++++++++++---------------------
 1 file changed, 15 insertions(+), 24 deletions(-)
drivers/usb/host/ehci-pci.c renamed from drivers/usb/host/ehci-pci.c~revert-usb-ehci-works-again-on-nvidia-controllers-with-2gb-ram +15 -24
diff -puN drivers/usb/host/ehci-pci.c~revert-usb-ehci-works-again-on-nvidia-controllers-with-2gb-ram drivers/usb/host/ehci-pci.c
@@ -76,30 +76,6 @@ static int ehci_pci_setup(struct usb_hcd
 	dbg_hcs_params(ehci, "reset");
 	dbg_hcc_params(ehci, "reset");
 
-        /* ehci_init() causes memory for DMA transfers to be
-         * allocated.  Thus, any vendor-specific workarounds based on
-         * limiting the type of memory used for DMA transfers must
-         * happen before ehci_init() is called. */
-	switch (pdev->vendor) {
-	case PCI_VENDOR_ID_NVIDIA:
-		/* NVidia reports that certain chips don't handle
-		 * QH, ITD, or SITD addresses above 2GB.  (But TD,
-		 * data buffer, and periodic schedule are normal.)
-		 */
-		switch (pdev->device) {
-		case 0x003c:	/* MCP04 */
-		case 0x005b:	/* CK804 */
-		case 0x00d8:	/* CK8 */
-		case 0x00e8:	/* CK8S */
-			if (pci_set_consistent_dma_mask(pdev,
-						DMA_31BIT_MASK) < 0)
-				ehci_warn(ehci, "can't enable NVidia "
-					"workaround for >2GB RAM\n");
-			break;
-		}
-		break;
-	}
-
 	/* cache this readonly data; minimize chip reads */
 	ehci->hcs_params = readl(&ehci->caps->hcs_params);
 
@@ -112,6 +88,8 @@ static int ehci_pci_setup(struct usb_hcd
 	if (retval)
 		return retval;
 
+	/* NOTE:  only the parts below this line are PCI-specific */
+
 	switch (pdev->vendor) {
 	case PCI_VENDOR_ID_TDI:
 		if (pdev->device == PCI_DEVICE_ID_TDI_EHCI) {
@@ -129,6 +107,19 @@ static int ehci_pci_setup(struct usb_hcd
 		break;
 	case PCI_VENDOR_ID_NVIDIA:
 		switch (pdev->device) {
+		/* NVidia reports that certain chips don't handle
+		 * QH, ITD, or SITD addresses above 2GB.  (But TD,
+		 * data buffer, and periodic schedule are normal.)
+		 */
+		case 0x003c:	/* MCP04 */
+		case 0x005b:	/* CK804 */
+		case 0x00d8:	/* CK8 */
+		case 0x00e8:	/* CK8S */
+			if (pci_set_consistent_dma_mask(pdev,
+						DMA_31BIT_MASK) < 0)
+				ehci_warn(ehci, "can't enable NVidia "
+					"workaround for >2GB RAM\n");
+			break;
 		/* Some NForce2 chips have problems with selective suspend;
 		 * fixed in newer silicon.
 		 */

_