Build issue on my Pi 5 'error: too many arguments to function ‘xsk_buff_dma_sync_for_cpu’'

[geerlingguy]

Here's the full log output:

pi@time-pi:~/intel-igc-ppsfix $ sudo dkms build --force igc -v 5.4.0-7642.46
Sign command: /lib/modules/6.12.62+rpt-rpi-2712/build/scripts/sign-file
Signing key: /var/lib/dkms/mok.key
Public certificate (MOK): /var/lib/dkms/mok.pub
Certificate or key are missing, generating self signed certificate for MOK...
Deprecated feature: REMAKE_INITRD (/var/lib/dkms/igc/5.4.0-7642.46/source/dkms.conf)

Building module:
Cleaning build area...
make -j4 KERNELRELEASE=6.12.62+rpt-rpi-2712 -C /lib/modules/6.12.62+rpt-rpi-2712/build M=/var/lib/dkms/igc/5.4.0-7642.46/build/src modules....(bad exit status: 2)
Error! Bad return status for module build on kernel: 6.12.62+rpt-rpi-2712 (aarch64)
Consult /var/lib/dkms/igc/5.4.0-7642.46/build/make.log for more information.
pi@time-pi:~/intel-igc-ppsfix $ cat /var/lib/dkms/igc/5.4.0-7642.46/build/make.log
DKMS make.log for igc-5.4.0-7642.46 for kernel 6.12.62+rpt-rpi-2712 (aarch64)
Wed 25 Feb 10:33:04 CST 2026
make: Entering directory '/usr/src/linux-headers-6.12.62+rpt-rpi-2712'
  CC [M]  /var/lib/dkms/igc/5.4.0-7642.46/build/src/igc_main.o
  CC [M]  /var/lib/dkms/igc/5.4.0-7642.46/build/src/igc_mac.o
  CC [M]  /var/lib/dkms/igc/5.4.0-7642.46/build/src/igc_i225.o
  CC [M]  /var/lib/dkms/igc/5.4.0-7642.46/build/src/igc_base.o
  CC [M]  /var/lib/dkms/igc/5.4.0-7642.46/build/src/igc_nvm.o
  CC [M]  /var/lib/dkms/igc/5.4.0-7642.46/build/src/igc_phy.o
  CC [M]  /var/lib/dkms/igc/5.4.0-7642.46/build/src/igc_diag.o
/var/lib/dkms/igc/5.4.0-7642.46/build/src/igc_main.c: In function ‘igc_construct_skb_zc’:
/var/lib/dkms/igc/5.4.0-7642.46/build/src/igc_main.c:2713:15: error: implicit declaration of function ‘__napi_alloc_skb’; did you mean ‘napi_alloc_skb’? [-Werror=implicit-function-declaration]
 2713 |         skb = __napi_alloc_skb(&ring->q_vector->napi, totalsize,
      |               ^~~~~~~~~~~~~~~~
      |               napi_alloc_skb
/var/lib/dkms/igc/5.4.0-7642.46/build/src/igc_main.c:2713:13: warning: assignment to ‘struct sk_buff *’ from ‘int’ makes pointer from integer without a cast [-Wint-conversion]
 2713 |         skb = __napi_alloc_skb(&ring->q_vector->napi, totalsize,
      |             ^
/var/lib/dkms/igc/5.4.0-7642.46/build/src/igc_main.c: In function ‘igc_clean_rx_irq_zc’:
/var/lib/dkms/igc/5.4.0-7642.46/build/src/igc_main.c:2816:17: error: too many arguments to function ‘xsk_buff_dma_sync_for_cpu’
 2816 |                 xsk_buff_dma_sync_for_cpu(bi->xdp, ring->xsk_pool);
      |                 ^~~~~~~~~~~~~~~~~~~~~~~~~
In file included from /var/lib/dkms/igc/5.4.0-7642.46/build/src/igc_main.c:13:
/usr/src/linux-headers-6.12.62+rpt-common-rpi/include/net/xdp_sock_drv.h:402:20: note: declared here
  402 | static inline void xsk_buff_dma_sync_for_cpu(struct xdp_buff *xdp)
      |                    ^~~~~~~~~~~~~~~~~~~~~~~~~
cc1: some warnings being treated as errors
make[2]: *** [/usr/src/linux-headers-6.12.62+rpt-common-rpi/scripts/Makefile.build:249: /var/lib/dkms/igc/5.4.0-7642.46/build/src/igc_main.o] Error 1
make[2]: *** Waiting for unfinished jobs....
make[1]: *** [/usr/src/linux-headers-6.12.62+rpt-common-rpi/Makefile:1970: /var/lib/dkms/igc/5.4.0-7642.46/build/src] Error 2
make: *** [/usr/src/linux-headers-6.12.62+rpt-common-rpi/Makefile:236: __sub-make] Error 2
make: Leaving directory '/usr/src/linux-headers-6.12.62+rpt-rpi-2712'

[geerlingguy]

Kernel:

pi@time-pi:~ $ uname -a
Linux time-pi 6.12.62+rpt-rpi-2712 #1 SMP PREEMPT Debian 1:6.12.62-1+rpt1~bookworm (2026-01-19) aarch64 GNU/Linux

[geerlingguy]

Only thing I see at a glance related to this is a 2024 patch: https://lkml.org/lkml/2024/5/6/390

[geerlingguy]

Also, this is on Bookworm; I may try out Trixie and see if there's a difference.

[JulianStj]

Added a new zip file for latest kernel, after doing full-upgrade on my rpi5 and using latest code, it all works. Please confirm before closing.

[geerlingguy]

Testing the newer version now:

pi@time-pi:~/intel-igc-ppsfix_6_12 $ sudo dkms add .
Deprecated feature: REMAKE_INITRD (/home/pi/intel-igc-ppsfix_6_12/dkms.conf)
Deprecated feature: REMAKE_INITRD (/usr/src/igc-6.12.0-ppsfix.1/dkms.conf)
Creating symlink /var/lib/dkms/igc/6.12.0-ppsfix.1/source -> /usr/src/igc-6.12.0-ppsfix.1
pi@time-pi:~/intel-igc-ppsfix_6_12 $ sudo dkms build --force igc -v 6.12.0-ppsfix.1
Sign command: /lib/modules/6.12.62+rpt-rpi-2712/build/scripts/sign-file
Signing key: /var/lib/dkms/mok.key
Public certificate (MOK): /var/lib/dkms/mok.pub
Deprecated feature: REMAKE_INITRD (/var/lib/dkms/igc/6.12.0-ppsfix.1/source/dkms.conf)

Building module:
Cleaning build area...
make -j4 KERNELRELEASE=6.12.62+rpt-rpi-2712 -C /lib/modules/6.12.62+rpt-rpi-2712/build M=/var/lib/dkms/igc/6.12.0-ppsfix.1/build/src modules.....
Signing module /var/lib/dkms/igc/6.12.0-ppsfix.1/build/src/igc.ko
Cleaning build area...
pi@time-pi:~/intel-igc-ppsfix_6_12 $ sudo dkms install --force igc -v 6.12.0-ppsfix.1
Deprecated feature: REMAKE_INITRD (/var/lib/dkms/igc/6.12.0-ppsfix.1/source/dkms.conf)

igc.ko.xz:
Running module version sanity check.
 - Original module
 - Installation
   - Installing to /lib/modules/6.12.62+rpt-rpi-2712/updates/dkms/
depmod...
pi@time-pi:~/intel-igc-ppsfix_6_12 $ sudo cp /lib/modules/$(uname -r)/kernel/drivers/net/ethernet/intel/igc/igc.ko.xz \
        /lib/modules/$(uname -r)/kernel/drivers/net/ethernet/intel/igc/igc.ko.xz.bak
pi@time-pi:~/intel-igc-ppsfix_6_12 $ sudo cp /lib/modules/$(uname -r)/updates/dkms/igc.ko.xz \
        /lib/modules/$(uname -r)/kernel/drivers/net/ethernet/intel/igc/igc.ko.xz
pi@time-pi:~/intel-igc-ppsfix_6_12 $ sudo depmod -a
pi@time-pi:~/intel-igc-ppsfix_6_12 $ sudo update-initramfs -u -k $(uname -r)
update-initramfs: Generating /boot/initrd.img-6.12.62+rpt-rpi-2712
'/boot/initrd.img-6.12.62+rpt-rpi-2712' -> '/boot/firmware/initramfs_2712'
pi@time-pi:~/intel-igc-ppsfix_6_12 $ sudo reboot

After reboot:

pi@time-pi:~ $ modinfo igc | grep filename
filename:       /lib/modules/6.12.62+rpt-rpi-2712/updates/dkms/igc.ko.xz
pi@time-pi:~ $ ethtool -i eth1
driver: igc
version: 6.12.62+rpt-rpi-2712
firmware-version: 2014:8877
expansion-rom-version: 
bus-info: 0001:01:00.0
supports-statistics: yes
supports-test: yes
supports-eeprom-access: yes
supports-register-dump: yes
supports-priv-flags: yes

Question: when verifying PPS, do I need to measure the output with an oscilloscope? And which SMA jack should I use?

# Read 5 external timestamps (should show ~1.000000000s intervals)
sudo testptp -d /dev/ptp0 -e 5

[geerlingguy]

Separate question: are the changes in igc upstreamed or have any potential of hitting mainline Linux at some point?

Also, do you have the actual patch file you're applying somewhere? Maybe we could use GitHub Actions to automatically build/zip new dkim versions?

[JulianStj]

You can measure PPS with an oscilloscope, but it will always look like 1PPS with or without this patch. To verify it with an oscilloscope, you'll need either a GPS 1PPS or another device with GPS locked 1PPS. The SMA port closer to the 40-pin connector is PPS output.

For upstreaming, that's outside my purview at the moment. If anybody has a contact or path for that, it'd be great.

I don't have a patch , just the files here. Shouldn't be too hard to make the patch, just grab the igc directory from the kernel matching this version and diff everything.

Will leave this issue open for further verification, thanks!

[geerlingguy]

You can measure PPS with an oscilloscope, but it will always look like 1PPS with or without this patch. To verify it with an oscilloscope, you'll need either a GPS 1PPS or another device with GPS locked 1PPS. The SMA port closer to the 40-pin connector is PPS output.

Okay, so for the verification I should have:

• Channel 1 getting PPS out of separate GPS
• Channel 2 getting PPS out of TimeHAT SMA plug nearest 40 pin header

Then I just confirm I'm getting rising edges within a few ns of each other? (Do you know about how accurate it should be, assuming everything's working, and I have a really good external reference?)

[JulianStj]

That's the easiest way of verifying yeah, after setting up ts2phc running and gps is locked on the TimeHAT.

Depending how good the GPS lock and GPS modules are, probably within 50-100 nanoseconds is to be expected, depending how long of a time window you look at for the min-max variation.