Arch Linux

this is now failling sha256 sum. when i get the checksum for the tarball it returns 36f7cf010a705b283f165e9dd611dac6ef4729c55466752c3b70f3ffcc6d47d2. anyone have any ideas?

I added EXTRA_FLAGS because it was complaining about TIME and TIME_DATE which was not in the makefile so I wanted to ignore them.

As to the patch file not being correct, what should I do as I'm new to this?

If it can't build to 3.16.1, what is the best course of action?

Where are you adding the EXTRA_CFLAGS to the Makefile? If I add -Wno-date-time to my CFLAGS, or to EXTRA_CFLAGS in the Makefile, it doesn't seem to make a difference. I can't build this version against 3.16.1, either way.

@minnowshark try this for your prepare:

prepare() {
cd "$srcdir/ceton_infinitv_linux_driver"
patch -p1 -i ../gcc.patch
}

However, it doesn't seem that what you pasted is a proper patch file:
==> Starting prepare()...
patching file Makefile
patch: **** unexpected end of file in patch

I'm getting an error trying to build.

----Start----

/home/mythtv/Downloads/ceton_infinitv/src/ceton_infinitv_linux_driver/ctn91xx_driver.c: In function ‘ctn91xx_print_compilation’:
/home/mythtv/Downloads/ceton_infinitv/src/ceton_infinitv_linux_driver/ctn91xx_driver.c:17:1: error: macro "__TIME__" might prevent reproducible builds [-Werror=date-time]
INFO("driver compiled at %s on %s", __TIME__, __DATE__);
-------

I'm new to this but I think I'm supposed to create a patch file and add extra flags. I don't know the structure but it should be something like this

-------

diff -Naur a/Makefile b/Makefile
--- a/Makefile 2013-08-01 08:52:22.000000000 +0200
+++ b/Makefile 2014-05-03 14:52:47.548492923 +0200
@@ -126,6 +126,8 @@

+EXTRA_CFLAGS += -Wno-date-time

-----------


Then add to the PKGBUILD file:


---------
prepare() {
cd "${pkgname}"
patch -p1 -i gcc.patch

}

-------

But I'm having problems with the directory. Any ideas?

broadcom-wl seem to have a relatively clean DKMS.conf and install file. Might be a good start.

I will give it a shot on Sunday and email you a copy if i get it working.

might be a possible just to modify one of the existing DKMS AUR packages.

shouldn't be too hard. currently don't have my ARCH running to test. but should have it up by Sunday.

I think I ran into a snag because I was trying to get it to compile based on the kernel headers that were installed instead of the running kernel. I'll go ahead, and throw you DKMS config in the package. Thanks for sharing it.

the DKMS config itself is pretty basic.

One issue i had is getting it to move the included udev rules.
Additionally, based on what i can find in a wiki it might have some conflicts with ABS which should be solvable with a symlink.

Here is my old DKMS config that was working below.

Granted it could be cleaned up a bit and it did work for me in arch when I manually added it to DKMS. Destination will probably need adjustments.



----------------------------START---------------------------------
MAKE="make -C ./ KERNELDIR=/lib/modules/${kernelver}/build"
CLEAN="make -C ./"
BUILD_MODULE_NAME=ctn91xx
DEST_MODULE_LOCATION=/kernel/modules/${kernelver}/extra/ceton
BUILD_MODULE_LOCATION=/ceton/
PACKAGE_NAME=ctn91xx
PACKAGE_VERSION=1.0
AUTOINSTALL=yes
REMAKE_INITRD=yes
-----------------------------END-------------------------------

I've tried setting up DKMS for it, but had some trouble. I'll look into it, again.

Would it make sense to make this install via DKMS? This way it would recompile automatically between minor kernel updates. I've done this before on Ubuntu and it seemed to work just fine across a bunch of kernel updates with no problems.

Ceton pushed the updates in http://cetoncorp.com/downloads/ceton_infinitv_linux_driver_2013_0326_2226.tar.gz

The reason for the rules patch is because when we look in /dev/ceton it does not exist with the updates. they end up in /dev/, therefore at the time the module gets loaded it creates a symlink for /dev/ceton and allows mythtv to run flawlessly.

We also need to patch the rules, these updates I've sent to the git and they have been committed.

--- 98-ctn91xx.rules 2012-11-16 14:19:50.000000000 -0600
+++ 98-ctn91xx.rules 2013-03-26 22:25:18.000000000 -0500
@@ -1 +1 @@
-KERNEL=="ctn91xx_*", NAME="ceton/%k", MODE="0666",OWNER="root",GROUP="root"
+KERNEL=="ctn91xx_*", SYMLINK+="ceton/%k", MODE="0666",OWNER="root",GROUP="root"

@tehsu, thanks for the patch. I hadn't tried to build this recently, but I'm updating it, now.

I fixed my issue, looks like its an issue with Kernel 3.8, heres my Patch.

--- ctn91xx_pci.old 2013-03-24 23:12:21.508354533 -0500
+++ ctn91xx_pci.c 2013-03-24 23:12:39.958648011 -0500
@@ -225,7 +225,7 @@
return -EIO;
}

-static void __devexit ctn91xx_unregister(struct pci_dev *pdev)
+static void ctn91xx_unregister(struct pci_dev *pdev)
{
ctn91xx_dev_t* dev = (ctn91xx_dev_t*)pci_get_drvdata(pdev);

@@ -264,7 +264,7 @@
kfree( dev );
}

-static struct pci_device_id ctn91xx_table[] __devinitdata = {
+static struct pci_device_id ctn91xx_table[] = {
{ CETON_VENDOR_ID, CTN91XX_DEVICE_ID, PCI_ANY_ID, PCI_ANY_ID, 0, 0, 0 },
{ ALTERA_VENDOR_ID, CTN91XX_DEVICE_ID, PCI_ANY_ID, PCI_ANY_ID, 0, 0, 0 },
{ 0, }
@@ -274,7 +274,7 @@
.name = DEVICE_NAME,
.id_table = ctn91xx_table,
.probe = ctn91xx_register,
- .remove = __devexit_p(ctn91xx_unregister),
+ .remove = ctn91xx_unregister,
};

int ctn91xx_register_pci_driver()

I'm seeing these errors, any ideas why?

/home/tehsu/Downloads/ceton_infinitv_linux_driver/ctn91xx_pci.c:228:23: error: expected ‘=’, ‘,’, ‘;’, ‘asm’ or ‘__attribute__’ before ‘ctn91xx_unregister’
/home/tehsu/Downloads/ceton_infinitv_linux_driver/ctn91xx_pci.c:267:45: error: expected ‘=’, ‘,’, ‘;’, ‘asm’ or ‘__attribute__’ before ‘__devinitdata’
/home/tehsu/Downloads/ceton_infinitv_linux_driver/ctn91xx_pci.c:275:17: error: ‘ctn91xx_table’ undeclared here (not in a function)
/home/tehsu/Downloads/ceton_infinitv_linux_driver/ctn91xx_pci.c:277:5: error: implicit declaration of function ‘__devexit_p’ [-Werror=implicit-function-declaration]
/home/tehsu/Downloads/ceton_infinitv_linux_driver/ctn91xx_pci.c:277:27: error: ‘ctn91xx_unregister’ undeclared here (not in a function)
/home/tehsu/Downloads/ceton_infinitv_linux_driver/ctn91xx_pci.c:290:1: error: ‘__mod_pci_device_table’ aliased to undefined symbol ‘ctn91xx_table’
cc1: some warnings being treated as errors

I haven't changed the location of modules since it doesn't break anything, but I'll make those changes for the next release.

Aside from the /usr/lib issue (which will be handled by symlink), the install path
/lib/modules/$KERNEL_VERSION/extra/
doesn't match my structure:
/lib/modules/3.7.7-1-ARCH/extramodules (where I find nVidia has been installed).
I'm guessing that this, too, is a non-issue, but haven't looked into it.
Does Linux not care about directory names for ko files?

Is the KERNEL_DIR variable used at all? I think it can be removed. Also, if you haven't planed for it yet, I'm pretty sure the next kernel release will have /lib/modules in /usr/lib/modules.

Please, update this, or disown it, so I can. Thanks.

Fixed for 3.1