Announcement

Collapse
No announcement yet.

Intel In-Tree Linux Network Drivers Being Adapted To Support Firmware Updates

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • Intel In-Tree Linux Network Drivers Being Adapted To Support Firmware Updates

    Phoronix: Intel In-Tree Linux Network Drivers Being Adapted To Support Firmware Updates

    Intel's IGB and IXGBE network drivers within the mainline Linux kernel are being adapted to support firmware updates for the underlying driver. To date such functionality was limited to Intel's out-of-tree versions of these drivers for their higher-end network hardware...

    Phoronix, Linux Hardware Reviews, Linux hardware benchmarks, Linux server benchmarks, Linux benchmarking, Desktop Linux, Linux performance, Open Source graphics, Linux How To, Ubuntu benchmarks, Ubuntu hardware, Phoronix Test Suite

  • #2
    Given how notoriously unstable Intel NICs are: https://www.youtube.com/watch?v=qB3FNijnrCk and that they can even brick their own firmware, and while flashing the firmware by users is not even supported by Intel and their board customers: https://www.youtube.com/watch?v=DXNyHFOWx_k. this is more than overdue. Next they also need to provide prosumer firmware to download, so users with self bricked NICs don't need to find them thru excessive internet search on obscure corners of the inter webs, ..! :-/

    Comment


    • #3
      Originally posted by rene View Post
      Given how notoriously unstable Intel NICs are: https://www.youtube.com/watch?v=qB3FNijnrCk and that they can even brick their own firmware, and while flashing the firmware by users is not even supported by Intel and their board customers: https://www.youtube.com/watch?v=DXNyHFOWx_k. this is more than overdue. Next they also need to provide prosumer firmware to download, so users with self bricked NICs don't need to find them thru excessive internet search on obscure corners of the inter webs, ..! :-/
      I cannot duplicate ANY of your claims, possibly because I do not have any Intel NICs that use the i225 chipset family.

      Issues in one chipset family, or perhaps it was just your own i225-based NICs "having a bad hair day & randomly bricking", is NOT CAUSE for concern that justifies a "POX on YOUR HOUSE" post that insinuates that ALL Intel NIC chipsets are bad.

      If all of those Intel chipsets going back multiple decades really were this bad then the news would be much more well known than a year old YT video by WCCF and another by some person name Rene'

      FWIW - I have used Intel NICs ever since 1Gbps switching devices became affordable for a home/consumer user. Before I started using Intel NICs I exclusively used 3Com NICs for 10/100 use.

      If any chipsets deserve a "POX on their house", IMHO, it would be Realtek 1Gpbs and slower; I do not have any faster networking than 1Gbps. From personal experience I have found that, side-by-side, my Realtek NICs cannot sustain the throughput that my Intel NICs can sustain. The Realtek NICs seem to just "barf & reset" or "barf & lock up" (requiring hard power reset) under heavy loads, like sustained 800-900Mbps file transfers through the same switching, and that is even with the latest Debian 12 OS and Realtek-firmware modules loaded.

      Comment


      • #4
        Originally posted by rene View Post
        Given how notoriously unstable Intel NICs are: https://www.youtube.com/watch?v=qB3FNijnrCk and that they can even brick their own firmware, and while flashing the firmware by users is not even supported by Intel and their board customers: https://www.youtube.com/watch?v=DXNyHFOWx_k. this is more than overdue. Next they also need to provide prosumer firmware to download, so users with self bricked NICs don't need to find them thru excessive internet search on obscure corners of the inter webs, ..! :-/
        That is complete and utter nonsense. Such lies.

        Comment


        • #5
          I did not make this YT video and spending 3h refreshing my NIC if it would not have broken itself overnight. Just look at the linux kernel git history of all the intel Nic drivers and all the quirks and hacks and disabling various acceleration, checksumming and other offloading. Most Intel NIC hw turned out to be so broken over the years that most drivers have all of those removed or disabled. Just google how many people have lost packets, link loss, low performance and other hiccup with each of the various intel ethernet implementations. I even quoted a news about intel acknowledging implementation bugs in that video: https://wccftech.com/intel-releases-...roller-issues/ https://www.tomshardware.com/news/in...around-for-now https://www.anandtech.com/show/18755...s-motherboards
          and only recently I realized the igc on my x670e board still stops working every 2nd week due to PCIe ASPM. Wasted another couple of hours over the last months to determine the "root cause", ...


          Feb 28 15:27:20 7950x kernel: igc 0000:0c:00.0 eth0: PCIe link lost, device now detached
          Feb 28 15:27:20 7950x kernel: ------------[ cut here ]------------
          Feb 28 15:27:20 7950x kernel: igc: Failed to read reg 0xc030!
          Feb 28 15:27:20 7950x kernel: WARNING: CPU: 2 PID: 2344 at drivers/net/ethernet/intel/igc/igc_main.c:6584 igc_update_stats+0x136/0x1f90 [igc]
          Feb 28 15:27:20 7950x kernel: Modules linked in: … igc …
          Feb 28 15:27:20 7950x kernel: CPU: 2 PID: 2344 Comm: kworker/2:2 Not tainted 6.5.6-t2 #1
          Feb 28 15:27:20 7950x kernel: Hardware name: Gigabyte Technology Co., Ltd. X670E AORUS MASTER/X670E AORUS MASTER, BIOS F22b 02/06/2024
          echo 0 > /sys/devices/pci0000:00/0000:00:02.1/0000:02:00.0/0000:03:08.0/0000:08:00.0/0000:09:06.0/0000:0c:00.0/link/l1_aspm
          Last edited by rene; 09 June 2024, 06:10 PM.

          Comment


          • #6
            Originally posted by brad0 View Post

            That is complete and utter nonsense. Such lies.
            just the whole internet and Intels own driver release and git history to make my point 🤡

            Comment


            • #7
              Originally posted by rene View Post
              just the whole internet and Intels own driver release and git history to make my point 🤡
              Overly dramatic nonsense, but what I would expect from a 🤡.

              Comment


              • #8
                The last time I tried to update firmware on an X550 NIC, it was on an Asrock board, and it resulted in the second port going dead. Apparently Asrock did something just wrong enough to make their package of Intel firmware work, but the stock Intel firmware not work. I had to flash back to their firmware by asking their support team for their secret special-sauce firmware package.

                Comment


                • #9
                  Originally posted by DanaG View Post
                  The last time I tried to update firmware on an X550 NIC, it was on an Asrock board, and it resulted in the second port going dead. Apparently Asrock did something just wrong enough to make their package of Intel firmware work, but the stock Intel firmware not work. I had to flash back to their firmware by asking their support team for their secret special-sauce firmware package.
                  Intel (like a number of other OEMs) sells their chipset to many mainboard and add-in board vendors. Some of those vendors choose to insure there are various lock-ins by using custom firmware and/or support chips which the Intel (or other OEM) firmware is unaware of and do not properly support. Sometimes cross flashing works, and sometimes it does not (just because the kernel says X710 (Intel example), or 2108 (LsiLogic example) one should never presume cross flashing will actually work without bricking the device).

                  Comment


                  • #10
                    I'm sorry, but I don't consider rene a troll or a clown. He may be controversial, but he usually provides solid facts.

                    He has his home lab and does youtube videos, but he also maintains a Linux distro to LOTS of platforms (many vintage servers and such).

                    He does many more, as he's an experienced developer woth both low level and high level programming from apps to drivers. He is the managing director of ExactCODE, practically the owner. He contrinutes to SANE. He also joined the TWAIN Working Group Board of Directors.

                    I understand he is weird and controversial but I agree tons with him after very carefully understanding and considering the facts he provide.

                    In the case of Intel, astroturfing and corporate marketing may brainwash us. Intel does bad hardware, but they have giant developer armies to apply workarounds massively. And after titanic efforts, it even may in some cases and more popular hardware to work better than a crappy realtek.

                    AMD has good hardware, bad software.
                    Intel has bad hardware, good software.
                    No company has both?

                    Why not AMD and Intel merge?
                    Last edited by timofonic; 10 June 2024, 12:05 AM.

                    Comment

                    Working...
                    X