Re: UEFI + SOL / COM ports = boot hang
The default state of this server is that FreeBSD hangs when booting off UEFI and the COM ports are enabled. That seems to be because console includes or adds "efi" by default. Under every circumstance I've tested, when console contains "efi", the server hangs at the same point shown in the first message. The workaround I've found is setting: console="comconsole" FreeBSD assumes that COM 1 is serial, which likely fails to initialize and so with only that setting, the video console remains primary. When that happens, I get no Beastie boot menu on either console. Since the server is now bootable with COM ports, I can tell the kernel where to find the serial port: # without the next 2 lines, the serial console isn't found and the video console remains primary comconsole_port="0x2F8" comconsole_speed="115200" And now it boots, I get the Beastie menu on the serial console, and later the FreeBSD kernel reports: Dual Console: Serial Primary, Video Secondary Changing the settings in BIOS related to console redirection seem not to matter much at all. Pretty much any combination of settings works. This isn't perfect. The UEFI video console is certainly nicer, and I'd prefer to have it be primary and serial be secondary. But I cannot include 'efi' in loader.conf console setting while the COM ports are enabled else FreeBSD hangs. The contents of the EFI variable ConOut with console ports disabled: # efivar -p -d 8be4df61-93ca-11d2-aa0d-00e098032b8c-ConOut 8be4df61-93ca-11d2-aa0d-00e098032b8c-ConOut : PciRoot(0x0)/Pci(0x1c,0x0)/Pci(0x0,0x0)/Pci(0x0,0x0)/AcpiAdr(0x80010100) With console ports enabled and console="comconsole": # efivar -p -d 8be4df61-93ca-11d2-aa0d-00e098032b8c-ConOut 8be4df61-93ca-11d2-aa0d-00e098032b8c-ConOut : PciRoot(0x0)/Pci(0x1c,0x0)/Pci(0x0,0x0)/Pci(0x0,0x0)/AcpiAdr(0x80010100),/PciRoot(0x0)/Pci(0x1f,0x0)/Serial(0x0)/Uart(115200,8,N,1)/UartFlowCtrl(Hardware)/VenUtf8(),/PciRoot(0x0)/Pci(0x1f,0x0)/Serial(0x1)/Uart(115200,8,N,1)/UartFlowCtrl(Hardware)/VenUtf8() From dmesg: EFI framebuffer information: addr, size 0x9c00, 0x30 dimensions 1024 x 768 stride 1024 masks 0x00ff, 0xff00, 0x00ff, 0xff00 VT(efifb): resolution 1024x768 ... vgapci0: port 0x2000-0x207f mem 0x9c00-0x9cff,0x9d00-0x9d01 irq 16 at device 0.0 numa-domain 0 on pci2 vgapci0: Boot video device I was hoping that setting hw.uart.console="io:0x2f8,br:115200" in loader.conf would help, but the server still hangs if 'efi' is included in the console list. Messing with hint.uart.N as suggested in https://wiki.freebsd.org/SerialConsole didn't help, I'm guessing that's because those settings aren't used by loader.efi. Any ideas or pointers for other settings which may enable the EFI console to be primary while also having a serial console? Thanks, Matt > On Sep 23, 2024, at 3:22 PM, Matt Simerson wrote: > > I have a Quanta D52B-1U server. When I boot it via legacy BIOS, there is no > issue. When I boot with UEFI and COM ports & console redirection disabled, > there is no issue. However, when boot UEFI and enable the COM ports and/or > console redirection, the server hangs at boot time right here: > > Version 2.20.1276. Copyright (C) 2023 American Megatrends, Inc.BIOS Date: > 12/04/2023 17:24:26 Ver: 3B28.Q102 > Consoles: EFI console > Reading loader env vars from /efi/freebsd/loader.env > Setting currdev to disk0p1: > FreeBSD/amd64 EFI loader, Revision 1.1 > >Command line arguments: loader.efi >Image base: 0x5e993000 >EFI version: 2.70 >EFI Firmware: American Megatrends (rev 5.14) >Console: efi (0x2000) >Load Path: \EFI\FREEBSD\LOADER.EFI >Load Device: > PciRoot(0x9)/Pci(0x0,0x0)/Pci(0x0,0x0)/NVMe(0x1,00-01-30-54-9D-E4-D2-5C)/HD(1,GPT,F669D424-7170-11EF-83D1-A0369F6BA978,0x28,0x82000) >BootCurrent: >BootOrder: [*] 0003 0006 0007 0002 0008 >BootInfo Path: > HD(1,GPT,F669D424-7170-11EF-83D1-A0369F6BA978,0x28,0x82000)/\EFI\FREEBSD\LOADER.EFI > Ignoring Boot: Only one DP found > Trying ESP: > PciRoot(0x9)/Pci(0x0,0x0)/Pci(0x0,0x0)/NVMe(0x1,00-01-30-54-9D-E4-D2-5C)/HD(1,GPT,F669D424-7170-11EF-83D1-A0369F6BA978,0x28,0x82000) > Setting currdev to disk0p1: > Trying: > PciRoot(0x9)/Pci(0x0,0x0)/Pci(0x0,0x0)/NVMe(0x1,00-01-30-54-9D-E4-D2-5C)/HD(2,GPT,4F537B49-7176-11EF-83D1-A0369F6BA978,0x82028,0x100) > Setting currdev to disk0p2: > Trying: > PciRoot(0x9)/Pci(0x0,0x0)/Pci(0x0,0x0)/NVMe(0x1,00-01-30-54-9D-E4-D2-5C)/HD(3,GPT,B6D8B384-7176-11EFLoading > /boot/defaults/loader.conf86860) > Loading /boot/defaults/loader.confdefault: > Loading /boot/device.hints > Loading /boot/loader.conf > Loading /boot/loader.conf.local > | > > On one occasion, I left it setting there for minutes. > > I currently have FreeBSD 14.1 installed and I also replicated this when > booting off a FreeBSD 15.0-CURRENT ISO. > > The chipset is Intel C621 & C624. Output from dmesg is here. > > BIOS
Re: September 2024 stabilization week
On Mon, Sep 23, 2024 at 01:00:11AM -0700, Gleb Smirnoff wrote: T> This is an automated email to inform you that the September 2024 stabilization week T> started with FreeBSD/main at main-n272449-6e414739fc95, which was tagged as T> main-stabweek-2024-Sep. The only issue discovered at Netflix testing is regression with ixl(4) 14561f1eda85 that has already been reverted with bf085659dac9. If you run with the main-stabweek-2024-Sep and use ixl(4) NIC you need to cherry-pick bf085659dac9. Other than that, no problems found. The advisory freeze on the main branch is thawed. -- Gleb Smirnoff
Re: New console output at boot?
On Wed, Sep 25, 2024 at 1:42 PM Thomas Laus wrote: > On 9/24/24 15:26, Warner Losh wrote: > > > > Thinking about it, I found a hole. If gptzfsboot picked the wrong BE > > this could happen. And updating it might fix a bug that's causing it. > > Knowing when your system last updated the gptzfsboot would be useful... > > > After getting a hint to use my cellphone camera to capture the error > messages, the errors all indicate missing fonts in /boot/fonts. How > does this directory get populated when building world? All of the fonts > in this directory date back to July 24, 2023 and nothing newer. Regarding the fonts I noticed that on 14.1-RELEASE AMD64 UEFI boot on laptop I have nice looking fonts, but on desktop I have old looking small resolution fonts.. so maybe something is on the thing with that fonts but I did not check :-P Also question is if serial console port can be used to catch the full bootloader process ? -- CeDeROM, SQ7MHZ, http://www.tomek.cedro.info
Re: New console output at boot?
Van: Thomas Laus Datum: woensdag, 25 september 2024 15:37 Aan: freebsd-current@freebsd.org Onderwerp: Re: New console output at boot? On 9/25/24 08:03, Ronald Klop wrote: > Hi, > > Nice work. I did the video trick also in the past with boot issues. > > Can you tell us what the errors are? Like what filenames are mentioned. > Maybe post the screenshot of the video somewhere. > > My /boot/fonts on raspberry pi 4 15.0-CURRENT contains. > $ ls -l /boot/fonts/ > total 187 > -r--r--r-- 1 root wheel 15355 Jul 7 2023 10x18.fnt.gz > -r--r--r-- 1 root wheel 15579 Jul 7 2023 10x20.fnt.gz > -r--r--r-- 1 root wheel 16772 Jul 7 2023 11x22.fnt.gz > -r--r--r-- 1 root wheel 17078 Jul 7 2023 12x24.fnt.gz > -r--r--r-- 1 root wheel 17158 Jul 7 2023 14x28.fnt.gz > -r--r--r-- 1 root wheel 20084 Jul 7 2023 16x32.fnt.gz > -r--r--r-- 1 root wheel 5874 Jul 7 2023 6x12.fnt.gz > -r--r--r-- 1 root wheel 12452 Jul 7 2023 8x14.fnt.gz > -r--r--r-- 1 root wheel 6472 Jul 7 2023 8x14v.fnt.gz > -r--r--r-- 1 root wheel 12604 Jul 7 2023 8x16.fnt.gz > -r--r--r-- 1 root wheel 6536 Jul 7 2023 8x16b.fnt.gz > -r--r--r-- 1 root wheel 6568 Jul 7 2023 8x16v.fnt.gz > -r--r--r-- 1 root wheel 2087 Aug 26 2023 INDEX.fonts > I have those exact fonts in my /boot/fonts/ that you do. The file sizes are the same but my dates are all July 24, 2023. It looks to me that either gptzfsboot should not look for some fonts or the buildworld process should install whatever gptzfsboot requires. My cellphone video only has this visible for a few frames and the entire video only lasts 5 seconds after typing in my GELI password until the boot menu is displayed. Tom -- Public Keys: PGP KeyID = 0x5F22FDC1 GnuPG KeyID = 0x620836CF I think it is still valuable to other developers if you could show us what the exact error messages are. Ronald.
Re: New console output at boot?
On 9/25/24 09:37, Thomas Laus wrote: On 9/25/24 08:03, Ronald Klop wrote: Hi, Nice work. I did the video trick also in the past with boot issues. Can you tell us what the errors are? Like what filenames are mentioned. Maybe post the screenshot of the video somewhere. My /boot/fonts on raspberry pi 4 15.0-CURRENT contains. $ ls -l /boot/fonts/ total 187 -r--r--r-- 1 root wheel 15355 Jul 7 2023 10x18.fnt.gz -r--r--r-- 1 root wheel 15579 Jul 7 2023 10x20.fnt.gz -r--r--r-- 1 root wheel 16772 Jul 7 2023 11x22.fnt.gz -r--r--r-- 1 root wheel 17078 Jul 7 2023 12x24.fnt.gz -r--r--r-- 1 root wheel 17158 Jul 7 2023 14x28.fnt.gz -r--r--r-- 1 root wheel 20084 Jul 7 2023 16x32.fnt.gz -r--r--r-- 1 root wheel 5874 Jul 7 2023 6x12.fnt.gz -r--r--r-- 1 root wheel 12452 Jul 7 2023 8x14.fnt.gz -r--r--r-- 1 root wheel 6472 Jul 7 2023 8x14v.fnt.gz -r--r--r-- 1 root wheel 12604 Jul 7 2023 8x16.fnt.gz -r--r--r-- 1 root wheel 6536 Jul 7 2023 8x16b.fnt.gz -r--r--r-- 1 root wheel 6568 Jul 7 2023 8x16v.fnt.gz -r--r--r-- 1 root wheel 2087 Aug 26 2023 INDEX.fonts I have those exact fonts in my /boot/fonts/ that you do. The file sizes are the same but my dates are all July 24, 2023. It looks to me that either gptzfsboot should not look for some fonts or the buildworld process should install whatever gptzfsboot requires. My cellphone video only has this visible for a few frames and the entire video only lasts 5 seconds after typing in my GELI password until the boot menu is displayed. Looking at the video again, I see gptzfsboot looking for most of these these same files without the .gz extension. Tom -- Public Keys: PGP KeyID = 0x5F22FDC1 GnuPG KeyID = 0x620836CF
Re: New console output at boot?
Hi, Nice work. I did the video trick also in the past with boot issues. Can you tell us what the errors are? Like what filenames are mentioned. Maybe post the screenshot of the video somewhere. My /boot/fonts on raspberry pi 4 15.0-CURRENT contains. $ ls -l /boot/fonts/ total 187 -r--r--r-- 1 root wheel 15355 Jul 7 2023 10x18.fnt.gz -r--r--r-- 1 root wheel 15579 Jul 7 2023 10x20.fnt.gz -r--r--r-- 1 root wheel 16772 Jul 7 2023 11x22.fnt.gz -r--r--r-- 1 root wheel 17078 Jul 7 2023 12x24.fnt.gz -r--r--r-- 1 root wheel 17158 Jul 7 2023 14x28.fnt.gz -r--r--r-- 1 root wheel 20084 Jul 7 2023 16x32.fnt.gz -r--r--r-- 1 root wheel 5874 Jul 7 2023 6x12.fnt.gz -r--r--r-- 1 root wheel 12452 Jul 7 2023 8x14.fnt.gz -r--r--r-- 1 root wheel 6472 Jul 7 2023 8x14v.fnt.gz -r--r--r-- 1 root wheel 12604 Jul 7 2023 8x16.fnt.gz -r--r--r-- 1 root wheel 6536 Jul 7 2023 8x16b.fnt.gz -r--r--r-- 1 root wheel 6568 Jul 7 2023 8x16v.fnt.gz -r--r--r-- 1 root wheel 2087 Aug 26 2023 INDEX.fonts An RPI3 running 14.1-RELEASE has the same filenames with the same sizes but with a date of May 2022. I didn't compare the content of the files. I'm not sure how they end up in that directory. In the FreeBSD sources I could only find share/syscons/fonts which contains some *.fnt files. I don't know if these are the same files as in /boot/fonts. Regards, Ronald. Van: Thomas Laus Datum: woensdag, 25 september 2024 13:42 Aan: freebsd-current@freebsd.org Onderwerp: Re: New console output at boot? On 9/24/24 15:26, Warner Losh wrote: > > Thinking about it, I found a hole. If gptzfsboot picked the wrong BE > this could happen. And updating it might fix a bug that's causing it. > Knowing when your system last updated the gptzfsboot would be useful... > After getting a hint to use my cellphone camera to capture the error messages, the errors all indicate missing fonts in /boot/fonts. How does this directory get populated when building world? All of the fonts in this directory date back to July 24, 2023 and nothing newer. Tom -- Public Keys: PGP KeyID = 0x5F22FDC1 GnuPG KeyID = 0x620836CF
Re: install error during installworld
On 25 Sep 2024, at 12:32, Gordon Bergling wrote: > > I have a few Hyper-V based virtual machines, one acts as a build machine > and one as -CURRENT development machine / environment. > > The host crashed while I was executing an installworld installkernel make > target. > > When trying to resume the installworld part I get the following error > message since then: > --- > make: "/storage/freebsd/src/current/share/mk/bsd.linker.mk" line 95: warning: > Unknown linker from LD=ld: , defaulting to bfd > find: : No such file or directory This is not good, my guess is that your linker was not found, therefore it printed something like "ld: not found" which then ended up being set in the LD make variable. Then bsd.linker.mk concludes that "ld:" is not a valid linker. Does /usr/bin/ld exist? Or /usr/local/bin/ld? > make[1]: warning: /storage/freebsd/src/current/: Read-only file system. > make[1]: > "/build/storage/freebsd/src/current/amd64.amd64/toolchain-metadata.mk" > line 1: Using cached toolchain metadata from build at bastion.ttyv0.de on Wed > Sep 25 12:20:39 CEST 2024 What is the contents of /build/storage/freebsd/src/current/amd64.amd64/toolchain-metadata.mk ? -Dimitry signature.asc Description: Message signed with OpenPGP
Re: New console output at boot?
On 9/24/24 15:26, Warner Losh wrote: Thinking about it, I found a hole. If gptzfsboot picked the wrong BE this could happen. And updating it might fix a bug that's causing it. Knowing when your system last updated the gptzfsboot would be useful... After getting a hint to use my cellphone camera to capture the error messages, the errors all indicate missing fonts in /boot/fonts. How does this directory get populated when building world? All of the fonts in this directory date back to July 24, 2023 and nothing newer. Tom -- Public Keys: PGP KeyID = 0x5F22FDC1 GnuPG KeyID = 0x620836CF
Re: New console output at boot?
On 9/25/24 08:03, Ronald Klop wrote: Hi, Nice work. I did the video trick also in the past with boot issues. Can you tell us what the errors are? Like what filenames are mentioned. Maybe post the screenshot of the video somewhere. My /boot/fonts on raspberry pi 4 15.0-CURRENT contains. $ ls -l /boot/fonts/ total 187 -r--r--r-- 1 root wheel 15355 Jul 7 2023 10x18.fnt.gz -r--r--r-- 1 root wheel 15579 Jul 7 2023 10x20.fnt.gz -r--r--r-- 1 root wheel 16772 Jul 7 2023 11x22.fnt.gz -r--r--r-- 1 root wheel 17078 Jul 7 2023 12x24.fnt.gz -r--r--r-- 1 root wheel 17158 Jul 7 2023 14x28.fnt.gz -r--r--r-- 1 root wheel 20084 Jul 7 2023 16x32.fnt.gz -r--r--r-- 1 root wheel 5874 Jul 7 2023 6x12.fnt.gz -r--r--r-- 1 root wheel 12452 Jul 7 2023 8x14.fnt.gz -r--r--r-- 1 root wheel 6472 Jul 7 2023 8x14v.fnt.gz -r--r--r-- 1 root wheel 12604 Jul 7 2023 8x16.fnt.gz -r--r--r-- 1 root wheel 6536 Jul 7 2023 8x16b.fnt.gz -r--r--r-- 1 root wheel 6568 Jul 7 2023 8x16v.fnt.gz -r--r--r-- 1 root wheel 2087 Aug 26 2023 INDEX.fonts I have those exact fonts in my /boot/fonts/ that you do. The file sizes are the same but my dates are all July 24, 2023. It looks to me that either gptzfsboot should not look for some fonts or the buildworld process should install whatever gptzfsboot requires. My cellphone video only has this visible for a few frames and the entire video only lasts 5 seconds after typing in my GELI password until the boot menu is displayed. Tom -- Public Keys: PGP KeyID = 0x5F22FDC1 GnuPG KeyID = 0x620836CF
install error during installworld
Hi, I have a few Hyper-V based virtual machines, one acts as a build machine and one as -CURRENT development machine / environment. The host crashed while I was executing an installworld installkernel make target. When trying to resume the installworld part I get the following error message since then: --- make: "/storage/freebsd/src/current/share/mk/bsd.linker.mk" line 95: warning: Unknown linker from LD=ld: , defaulting to bfd find: : No such file or directory make[1]: warning: /storage/freebsd/src/current/: Read-only file system. make[1]: "/build/storage/freebsd/src/current/amd64.amd64/toolchain-metadata.mk" line 1: Using cached toolchain metadata from build at bastion.ttyv0.de on Wed Sep 25 12:20:39 CEST 2024 -- >>> Install check world started on Wed Sep 25 10:26:40 UTC 2024 -- usage: mkdir [-pv] [-m mode] directory_name ... --- installworld --- *** [installworld] Error code 64 make[1]: stopped making "installworld" in /storage/freebsd/src/current make[1]: 1 error make[1]: stopped making "installworld" in /storage/freebsd/src/current --- installworld --- make: stopped making "installworld" in /storage/freebsd/src/current --- I deleted the obj directory and did a fresh build, but the error persists. Somewhat the development VM seems to be corrupt. Has any one an idea how to solve this problem? Any help is much appreciated. --Gordon signature.asc Description: PGP signature
