Opened 4 years ago
Last modified 3 years ago
#16358 new bug
Boot from CD Fails With "PANIC: did not find any boot partitions!"
Reported by: | petal | Owned by: | nobody |
---|---|---|---|
Priority: | normal | Milestone: | Unscheduled |
Component: | Drivers/Disk/ATA | Version: | R1/Development |
Keywords: | boot-failure | Cc: | |
Blocked By: | #6558 | Blocking: | |
Platform: | All |
Description
Summary
When booting from CD on my x86 system, all proceeds as expected through the fifth icon, at which point a kernel panic occurs with the error PANIC: did not find any boot partitions!
Tested on both R1/beta2 and Nightly hrev54384-x86_gcc2h.
Details
The system is a Compaq Presario 3000 (Model: 3045us) Pentium 4 laptop (an ancient machine, circa 2003; the best remaining online source for some specs is available here: https://www.cnet.com/products/hp-compaq-presario-3045us/specs/).
It is a stock machine, except for the HDD which has been replaced with a generic PATA-to-mSATA board and a 256GiB mSATA drive; and an upgrade to 1GiB of PC3200 RAM. The conversion board board (like many of these conversion boards, I'm told) doas not handle DMA correctly. Booting with IDE DMA disabled did not fix the problem. If it helps to inspire any confidence, this board has worked with other operating systems with IDE DMA disabled.
Various combinations of debugging options (Safe mode, Disable IDE DMA, Disable ACPI, etc.) did not get me any further.
Attachments
My apologies for the the pic attachments. Even with a USB drive plugged in, no "Display syslog from previous session" or "Save syslog from previous session" message appeared in the debug menu. The attached pics are of the panic message, and the last few screens of the syslog. I would be happy to photograph more and send it along if it would help. (At 640x480, the syslog is about 20 screens long).
Attachments (30)
Change History (44)
by , 4 years ago
Attachment: | IMG_20200704_122232.jpg added |
---|
comment:1 by , 4 years ago
Component: | - General → Drivers/Disk/ATA |
---|---|
Platform: | x86 → All |
comment:2 by , 4 years ago
It's a bit odd, as the last part of the syslog clearly shows that the right partition was found. Offsets and sizes look sane, so the actual error has to be earlier. Can you attach screenshots from before the very verbose disk dump starts?
Can you also check whether or not the boot volume shows up in the boot loader menu and try selecting it explicitly?
Regarding the "previous syslog", did you reboot via button or by running the "reboot" command in KDL? The latter should soft reset the machine so that the debug syslog remains intact.
comment:3 by , 4 years ago
Please find the requested pics attached. I started taking photos when ACPI and usb messages started showing up; please let me know if I should still go further back.
No boot volume shows up in the boot loader menu. My apologies for not saying so earlier: the disk is entirely blank (formatted "cleared" with GParted) with a 1MB offset from the start of the disk. This error is entirely pre-installation. (The same result occurs when formatted with "unformatted," and whether the boot flag is active or not seems to make no difference).
My reboots have all been via "reboot" in KDL, with no luck, I'm sorry to say =/.
comment:4 by , 4 years ago
The boot message dump tells that it uses the hard disk boot method, which is unexpected. As you're running from a CD, the CD boot method should be used. This affects how the boot device and boot volume are selected and probably leads to this. I have not yet checked how the device type is detected, but I would expect the BIOS to report the CD as a hard disk, possibly because it has an MBR due to the anyboot layout.
comment:7 by , 4 years ago
The syslog there shows that it got a boot method CD, so this seems different.
In both cases the boot partition is indeed readable and identified properly, but then not used. In this case, due to the wrong boot method, the selection is probably too strict.
One could argue that in both cases it should maybe try harder and boot the most probable target if none can be exactly determined. There is only a single possible boot source in this case. This would then possibly lead to a case where firmware is able to read a volume and make it available to the bootloader that we don't actually have a driver for. In such a case we would then maybe boot a wrong partition. This would probably still be better than failing also in straight forward cases like the one here.
comment:8 by , 4 years ago
Can you retry with >= hrev54418? This might be affected by the PCI 64 bit detection bug that caused IO ports to be skipped.
From the first "middle" picture it also shows all USB controllers failing to init and a PCI read failure. So there may be a more fundamental PCI init issue at hand.
comment:9 by , 4 years ago
Sorry for the delay! It took me a bit to find some time to give this a go.
The bottom line
Booting still fails with the same error message.
No USB stick is detected, and the option to save the previous syslog is not displayed in the debug options of the boot loader options menu. So... a series of syslog pics will again be attached. I'm sorry for this!
Some Details
- In the meantime, I've installed another OS on this system, so there's no empty partition to install to. I assume this shouldn't make a difference, but I thought it be worth mentioning.
- These boot attempts were made on hrev54418-x86_gcc2h
- Attepmts were made both as-is, and with the "Safe Mode" and "Disable IDE DMA" boot options at 640x480 switched on.
- A complete syslog in pics will be attached following this message.
Thanks for your continuing help! Please let me know if I'm forgetting anything or can contribute further.
comment:10 by , 4 years ago
Please do not attach all 20 syslogs here. If you can, make an Imgur album with them or upload them elsewhere, and we can then select only the relevant ones.
comment:11 by , 4 years ago
I've finally been able to look through the pictures here and at least why the USB stick isn't seen in the bootloader and USB init fails is now clear: The firmware apparently doesn't initialize the USB controllers on the PCI side. Their interrupt configuration and BARs are all unconfigured, so they are unusable (due to #3 and #5). Can you check if there's a USB legacy support or "boot from USB" setting to turn on in the BIOS? Might also be a "PnP OS installed" setting that you could turn off. If that'd work you could possibly try a USB boot instead of using the CD.
There are no other errors or indications that something else is going wrong, so to me the CD boot issue still looks like one of having the wrong boot method and therefore disregarding the valid looking CD.
comment:12 by , 4 years ago
Thanks once again for following up on this! Unfortunately, this laptop is *just* a bit too old (circa 2003) for a Boot from USB option (in fact, it came bundled with the first flash drive I'd ever seen, at a whopping 16MB!); only booting from Floppy, Hard Drive, or CD are supported.
There *is* a Legacy USB option in the BIOS, but the stick is not detected for logging purposes whether this option is enabled or not.
No "PnP OS installed" option is available.
So, unfortunately, I'm stuck with CD for the time being =].
comment:13 by , 4 years ago
IIRC there is a Plop Boot Manager (https://www.plop.at/en/bootmanager/usbinfo.html) that can be used to boot from USB drives even if BIOS doesn't support USB booting.
comment:14 by , 3 years ago
Keywords: | boot-failure added |
---|
Kernel Panic Message