Bug 258319

Summary: Network (em0) does'nt work not only in 13.0-RELEASE but also in it's installer.
Product: Base System Reporter: nctto14935
Component: kernAssignee: freebsd-net (Nobody) <net>
Status: New ---    
Severity: Affects Some People CC: grahamperrin, kbowling, linimon
Priority: --- Keywords: install, regression
Version: 13.0-RELEASE   
Hardware: i386   
OS: Any   

Description nctto14935 2021-09-06 16:45:49 UTC
In 11.4-RELESE/i386 or before, network (em0, IPv4) works well (in base system as well as in it's installer).  But in 13.0-RELEASE/i386, network does'nt work not only in the base system but also in it's installer. I have confirmed this by three different methods as follows:

(1) Updating 11.4-RELEASE/i386 to 13.0-RELEASE/i386
freebsd-update works perfectly and it successfully upgrades 11.4 (of which network works well) into 13.0, but network suddenly stops just after rebooting the system with 13.0 kernel.
(2) Network installation of 13.0-RELEASE/i386
The bootonly installer of 13.0-RELEASE works halfway.  It fails to fetch 13.0-RELEASE files from the internet (since network does'nt work) and finally ends with error messages.  This is a problem of 13.0 installer, but I feel this problem stems from the same origin. 
(3) DVD Installaton of 13.0-RELEASE/i386
The dvd1 installer of 13.0-RELEASE successfully installs 13.0-RELEASE, but after rebooting the installed system, one finds network does'nt work.

In addition I have tried 12.2-RELEASE/i386 bootonly installer, and found that 12.2 installer shares the same problem. (In contrast, 11.4 installer or before works perfectly.)

In 13.0-RELEASE/i386, network phenomena as follows are occuring:

(1) "ping hostname" ends with "Unknown host".
(2) "ping IP-address-of-host" works well.
(3) "ping localhost" works well.
(4) "ping IP-address-of-DNS-server" ends with "Host is down".
(5) "ping IP-address-of-gateway" ends with "Host is down"
(6) "ssh IP-address-of-host" from another computer fails.

I have confirmed these phenomena just after the clean installation of 13.0-RELEASE/i386.  According to dmesg messages, 13.0 kernel seems to recognize em0.  As for 13.0-RELEASE/amd64, I have never encountered the network problem like this.
Comment 1 Mark Linimon freebsd_committer freebsd_triage 2021-09-07 21:24:06 UTC
Please show us the output of:

  dmesg | grep em0

Also please the content of the block from pciconf -lv that has the line:

  subclass   = ethernet

Thanks.
Comment 2 nctto14935 2021-09-11 14:38:05 UTC
# dmesg | grep em0
em0: <Intel(R) PRO/1000 Network Connection> port 0xac00-0xac1f mem 0xff7e0000-0xff7fffff irq 18 at device 1.0 on pci1
em0: Using 1024 TX descriptors and 1024 RX descriptors
em0: Using a Legacy interrupt
em0: Ethernet address: xx:xx:xx:xx:xx:xx
em0: link state changed to UP

# pciconf -lv
em0@pci0:1:1:0: class=0x020000 rev=0x00 hdr=0x00 vendor=0x8086 device=0x1019 subvendor=0x8086 subdevice=0x303b
    vendor     = 'Intel Corporation'
    device     = '82547EI Gigabit Ethernet Controller'
    class      = network
    subclass   = ethernet
r10@pci0:2:0:0: class=0x020000 rev=0x10 hdr=0x00 vendor=0x10ec device=0x8139 subvendor=0x10fc subdevice=0xd01b
    vendor     = 'Realtek Semiconductor Co., Ltd.'
    device     = 'RTL-8100/8101L/8139 PCI Fast Ethernet Adapter'
    class      = network
    subclass   = ethernet