Bug#670241: Updated qmi_wwan backport based on v3.2.19, including new device IDs from v3.5-rc1
Ben Hutchings writes: > On Tue, Jul 17, 2012 at 08:01:24PM +0200, Bjørn Mork wrote: > >> Feel free to close this bug with a wontfix if you like. After all, the >> qmi_wwan driver will never be very useful in wheezy, given that the >> modemmanager package is frozen on a version too old to ever gain QMI >> support. > > How large are the required changes to ModemManager? Huge. The next version (0.6) will use a completely new D-Bus API making it incompatible with the current NetworkManager and any other ModemManager users. Ref http://www.freedesktop.org/wiki/ModemManager06 And the QMI support might even not be ready for 0.6. It's currently being developed in a separate branch and not yet fully operational. > New hardware > support is a perfectly good reason for a freeze exception, whichever > package it's in. Sure, and I appreciate that. But do note that this driver does not add support for any new USB device. All supported modems are composite devices having at least one serial port supporting PPP. This means that the driver isn't critical for basic device support. It only adds support for another device function, which most users will see as redundant. Bjørn -- To UNSUBSCRIBE, email to debian-kernel-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/87pq7ujj48@nemi.mork.no
Bug#670241: Updated qmi_wwan backport based on v3.2.19, including new device IDs from v3.5-rc1
On Tue, Jul 17, 2012 at 08:01:24PM +0200, Bjørn Mork wrote: > block 670241 by 681912 > thanks > > Please don't implement this just yet. The ModemManager version in > wheezy may choke on the ports provided by the new driver. Ref > https://bugzilla.redhat.com/show_bug.cgi?id=835153 > > I've opened a bug against ModemManager requesting the addition of the > upstream workaround, and am now blocking this bug against it. Better > not include the driver until we sure it won't trigger any ModemManager > regressions. > > Feel free to close this bug with a wontfix if you like. After all, the > qmi_wwan driver will never be very useful in wheezy, given that the > modemmanager package is frozen on a version too old to ever gain QMI > support. How large are the required changes to ModemManager? New hardware support is a perfectly good reason for a freeze exception, whichever package it's in. Ben. > And thanks for delaying this until Fedora sorted out the worst problems > :-) That's wise. -- Ben Hutchings We get into the habit of living before acquiring the habit of thinking. - Albert Camus -- To UNSUBSCRIBE, email to debian-kernel-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/20120717183748.gw1...@decadent.org.uk
Processed: Re: Bug#670241: Updated qmi_wwan backport based on v3.2.19, including new device IDs from v3.5-rc1
Processing commands for cont...@bugs.debian.org: > block 670241 by 681912 Bug #670241 [linux-2.6] linux-2.6: Backporting the qmi_wwan driver to the Debian 3.2/wheezy kernels 670241 was not blocked by any bugs. 670241 was not blocking any bugs. Added blocking bug(s) of 670241: 681912 > thanks Stopping processing here. Please contact me if you need assistance. -- 670241: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=670241 Debian Bug Tracking System Contact ow...@bugs.debian.org with problems -- To UNSUBSCRIBE, email to debian-kernel-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/handler.s.c.13425484859791.transcr...@bugs.debian.org
Bug#670241: Updated qmi_wwan backport based on v3.2.19, including new device IDs from v3.5-rc1
block 670241 by 681912 thanks Please don't implement this just yet. The ModemManager version in wheezy may choke on the ports provided by the new driver. Ref https://bugzilla.redhat.com/show_bug.cgi?id=835153 I've opened a bug against ModemManager requesting the addition of the upstream workaround, and am now blocking this bug against it. Better not include the driver until we sure it won't trigger any ModemManager regressions. Feel free to close this bug with a wontfix if you like. After all, the qmi_wwan driver will never be very useful in wheezy, given that the modemmanager package is frozen on a version too old to ever gain QMI support. And thanks for delaying this until Fedora sorted out the worst problems :-) That's wise. Bjørn -- To UNSUBSCRIBE, email to debian-kernel-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/87txx6jmh7@nemi.mork.no
Bug#670241: Updated qmi_wwan backport based on v3.2.19, including new device IDs from v3.5-rc1
Ben Hutchings writes: > On Mon, 2012-06-04 at 22:07 +0200, Bjørn Mork wrote: >> A few changes since the bug was reported: >> >> - v3.2.19 includes a few fixes in cdc-wdm which conflict with the >> backport. The easiest way to merge these is to first revert them, do >> the backport, and then reapply the fixes as cherry-picks from >> mainline. > > Having to revert is not ideal, but it's OK. The advantage is that all mainline patches apply cleanly. I can prepare a set without the reverts if you want that, but that would mean a number of context adjustments in the backport. > [...] >> The resulting drivers will be identical to the 3.5-rc1 versions, with >> the exception of the module_usb_driver() macro and the >> disable_hub_initiated_lpm patch: > [...] > > I'm quite happy to add module_usb_driver() if it helps to make cleaner > backports. I don't think it helps much. The mainline patch applying the macro to the drivers cannot be used anyway, as that is a bulk change touching a large number of drivers at once. And this part of the driver is rarely used as context, so it doesn't affect backporting other patches. Better just leave it as it is. Bjørn -- To UNSUBSCRIBE, email to debian-kernel-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/87zk8iuql6@nemi.mork.no
Bug#670241: Updated qmi_wwan backport based on v3.2.19, including new device IDs from v3.5-rc1
On Mon, 2012-06-04 at 22:07 +0200, Bjørn Mork wrote: > A few changes since the bug was reported: > > - v3.2.19 includes a few fixes in cdc-wdm which conflict with the > backport. The easiest way to merge these is to first revert them, do > the backport, and then reapply the fixes as cherry-picks from > mainline. Having to revert is not ideal, but it's OK. [...] > The resulting drivers will be identical to the 3.5-rc1 versions, with > the exception of the module_usb_driver() macro and the > disable_hub_initiated_lpm patch: [...] I'm quite happy to add module_usb_driver() if it helps to make cleaner backports. Should be just these two: commit 907d0ed1c84114d4e8dafd66af982515d3739c90 Author: Lars-Peter Clausen Date: Wed Nov 16 10:13:35 2011 +0100 drivercore: Generalize module_platform_driver commit f3a6a4b6cfc80e57bf16bb12f9425bec1a5731a9 Author: Greg Kroah-Hartman Date: Thu Nov 17 14:38:33 2011 -0800 USB: Add helper macro for usb_driver boilerplate Ben. -- Ben Hutchings It is impossible to make anything foolproof because fools are so ingenious. signature.asc Description: This is a digitally signed message part
Bug#670241: Updated qmi_wwan backport based on v3.2.19, including new device IDs from v3.5-rc1
A few changes since the bug was reported: - v3.2.19 includes a few fixes in cdc-wdm which conflict with the backport. The easiest way to merge these is to first revert them, do the backport, and then reapply the fixes as cherry-picks from mainline. - The sierra driver fix has been included in stable and can be dropped. - A few new device IDs has been added to the qmi_wwan driver. The following procedure will produce a backported version of qmi_wwan and the supporting cdc-wdm driver on top of a clean v3.2.19: 1) revert the following conflicting stable fixes (will be reapplied in the next step): 8fd4242 USB: cdc-wdm: cannot use dev_printk when device is gone 0c68ab1 USB: cdc-wdm: add debug messages on cleanup 7569109 USB: cdc-wdm: fix memory leak 811c72b USB: cdc-wdm: sanitize error returns 2) cherry-pick to add subdriver support to cdc-wdm: 19b85b3 USB: cdc-wdm: no need to fill the in request URB every time it's submitted 8143a89 USB: cdc-wdm: kill the now unnecessary bMaxPacketSize0 field and udev variable 820c629 USB: cdc-wdm: avoid printing odd-looking "cdc-wdm-176" names fec67b4 usb: cdc-wdm: Add device-id for Huawei 3G/LTE modems 8804420 usb: cdc-wdm: make reset work with blocking IO 8457d99 USB: cdc-wdm: no need to use usb_alloc_coherent 0dffb48 usb: cdc-wdm: split out reusable parts of probe b0c1386 usb: cdc-wdm: adding list lookup indirection 3cc3615 usb: cdc-wdm: adding usb_cdc_wdm_register subdriver support 24a85ba USB: cdc-wdm: sanitize error returns 2f338c8 USB: cdc-wdm: fix memory leak 880bca3 USB: cdc-wdm: add debug messages on cleanup 12a98b2 USB: cdc-wdm: cleanup error codes 6b0b79d USB: cdc-wdm: cannot use dev_printk when device is gone 6286d85 USB: cdc-wdm: remove from device list on disconnect 3) cherry-pick to add qmi_wwan driver: 423ce8c net: usb: qmi_wwan: New driver for Huawei QMI based WWAN devices c3ecb08 net: qmi_wwan: support devices having a shared QMI/wwan interface b086cf0 net: qmi_wwan: add Gobi and Pantech UML290 device IDs 11207b6 net: qmi_wwan: add support for ZTE MF820D 1aa35a2 USB: qmi_wwan: Add ZTE (Vodafone) K3565-Z and K4505-Z net interfaces dbb6d09 USB: qmi_wwan: Add ZTE (Vodafone) K3570-Z and K3571-Z net interfaces 3bc17d1 net: qmi_wwan: support Sierra Wireless MC77xx devices in QMI mode 88c16dc net: qmi_wwan: Add Vodafone/Huawei K5005 support 188 USB: qmi_wwan: Make forced int 4 whitelist generic 8965c98 USB: qmi_wwan: Add ZTE (Vodafone) K3765-Z f7142e6 USB: qmi_wwan: Add ZTE (Vodafone) K3520-Z 5e071b5 net: qmi_wwan: Add Sierra Wireless device IDs The diffstat of the above is: bjorn@nemi:/usr/local/src/git/linux$ git diff --stat v3.2.19..HEAD drivers/net/usb/Kconfig | 22 +++ drivers/net/usb/Makefile|1 + drivers/net/usb/qmi_wwan.c | 581 drivers/usb/class/cdc-wdm.c | 345 include/linux/usb/cdc-wdm.h | 19 ++ 5 files changed, 849 insertions(+), 119 deletions(-) The resulting drivers will be identical to the 3.5-rc1 versions, with the exception of the module_usb_driver() macro and the disable_hub_initiated_lpm patch: bjorn@nemi:/usr/local/src/git/linux$ git diff HEAD..v3.5-rc1 -- drivers/usb/class/cdc-wdm.c include/linux/usb/cdc-wdm.h drivers/net/usb/qmi_wwan.c diff --git a/drivers/net/usb/qmi_wwan.c b/drivers/net/usb/qmi_wwan.c index 32755c6..3b20678 100644 --- a/drivers/net/usb/qmi_wwan.c +++ b/drivers/net/usb/qmi_wwan.c @@ -562,6 +562,7 @@ static struct usb_driver qmi_wwan_driver = { .resume = qmi_wwan_resume, .reset_resume = qmi_wwan_resume, .supports_autosuspend = 1, + .disable_hub_initiated_lpm = 1, }; static int __init qmi_wwan_init(void) diff --git a/drivers/usb/class/cdc-wdm.c b/drivers/usb/class/cdc-wdm.c index 2236fe7..ea8b304 100644 --- a/drivers/usb/class/cdc-wdm.c +++ b/drivers/usb/class/cdc-wdm.c @@ -1034,26 +1034,10 @@ static struct usb_driver wdm_driver = { .post_reset = wdm_post_reset, .id_table = wdm_ids, .supports_autosuspend = 1, + .disable_hub_initiated_lpm = 1, }; -/* --- low level module stuff --- */ - -static int __init wdm_init(void) -{ - int rv; - - rv = usb_register(&wdm_driver); - - return rv; -} - -static void __exit wdm_exit(void) -{ - usb_deregister(&wdm_driver); -} - -module_init(wdm_init); -module_exit(wdm_exit); +module_usb_driver(wdm_driver); MODULE_AUTHOR(DRIVER_AUTHOR); MODULE_DESCRIPTION(DRIVER_DESC); Bjørn -- To UNSUBSCRIBE, email to debian-kernel-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/87fwaax2lr@nemi.mork.no