PANIC: get_boot_partitions failed

[AlexeiZavjalov]

Haiku revision: hrev47127

Serial dump is attached.

[bonefish]

I'm not particularly familiar with the AHCI module, but the output in the log suggests that something goes wrong there:

1207	ahci: AHCIPort::ScsiTestUnitReady port 0
1208	ahci: AHCIPort::ScsiInquiry port 0
1209	ahci: ExecuteAtaRequest port 0: device is busy
1210	ahci: AHCIPort::ResetPort port 0
1211	ahci: AHCIPort::ResetPort port 0, deviceBusy 1, forceDeviceReset 0
1212	ahci: AHCIPort::PostReset port 0
1213	ahci: device signature 0x00000101 (ATA)
1214	ahci: sata_request::abort called for command 0xec
1215	ahci: identify device failed
1216	ahci: ExecuteAtaRequest port 1: device is busy
1217	ahci: sata_request::abort called for command 0xa0
1218	ahci: ExecuteAtaRequest port 1: device is busy
1219	ahci: AHCIPort::ResetPort port 1
1220	ahci: AHCIPort::ResetPort port 1, deviceBusy 1, forceDeviceReset 0
1221	ahci: AHCIPort::PostReset port 1
1222	ahci: device signature 0xeb140101 (ATAPI)
1223	ahci: sata_request::abort called for command 0xa0

[AlexeiZavjalov]

The same problem I have on the R1/Alpha 4.1 release

[AlexeiZavjalov]

I found that ATAChannel::_DevicePresent expects SelectedDevice() will return 0, but it returns 1.

The problem has gone after I turned off the SATA mode.

[umccullough]

It seems to me that no x86_64 issues should be a blocker for any R1 related release, since it's totally experimental and unofficial still.

[umccullough]

Ah, nevermind, the issue is present on Alpha 4.1 it seems, changing the summary.

[waddlesplash]

Related to #12295?

[waddlesplash]

There have been a lot of AHCI fixes in the past 3 years, including #12295 from two years ago. Could you please retest?

[rhodes]

Dropped into KDL with the same error (PANIC: get_boot_partitions failed!) on a Samsung Chromebook 2 (model xe500c12) booting haiku-nightly-hrev51528-x86_64-anyboot from USB.

The same panic occurs using the haiku-nightly-hrev51536-x86_gcc2_hybrid-anyboot image.

[rhodes]

chromebook xe500c12 KDL

[waddlesplash]

Between hrev51528 and hrev51536 there were some major fixes made to the XHCI bus manager ... so are you sure there are no differences between the two in the KDL message?

[rhodes]

Here's the KDL from hrev51538

[waddlesplash]

Can you enable onscreen debug logging in the bootloader to capture that output, or even use the "dump last syslog" function following a warm reset? (see ​https://www.haiku-os.org/docs/userguide/en/bootloader.html)

[rhodes]

Syslog from hrev51538: ​https://pastebin.com/5PPRJVpF

[waddlesplash]

ata 0: controller doesn't support DMA, disabling
ata 0: _DevicePresent: device selection failed for device 0
ata 0: _DevicePresent: device 1, presence 0
ata 0: deviceMask 0
ata 0: ignoring device 0
ata 0: ignoring device 1
ata 0 error: target device not present
ata 0 error: target device not present
ata 0 error: invalid target device

Huh, never seen that error before. Interesting.

[waddlesplash]

The panic in that final syslog is unrelated.

The ATA error I snipped in my comment above is also unrelated, as obviously an ATA error could not prevent a boot from USB, and I've actually seen that error on systems where booting from ATA works just fine.

So, it seems this ticket is actually no longer reproducible.

[nielx]

Remove milestone for tickets with status = closed and resolution != fixed