Custom Query (16363 matches)
Results (607 - 609 of 16363)
| Ticket | Resolution | Summary | Owner | Reporter |
|---|---|---|---|---|
| #14008 | invalid | AHCI Port Reset After 1500 Hours Without Any Warnings | ||
| Description |
Dear VirtualBox community. Our guest suddenly stop working properly after 1500 hours . First we cannot ssh into it although it is still responsive to ping request. After restart we found that guest files and /var/log/messages for the last 5 hours did not written to disk. VBox.log shows that an AHCI port reset occured 5 hours beforehand. By accident our system administrator restart the host instead of the guest, that is why there are no other information in VBox.log.1 after port reset. Unlike bug #9975 this reset did not precede by any warnings nor errors. We make a guess that this is an hard reset and from what we learn from AHCI specification, hard reset is triggered by software, which in this case is linux driver. What we don't know is why the driver reset the port. Is this because something in VD/VDI or AHCI emulation ? Please help us. Another information the host disk is configured as raid-5. Both Host and Guest are linux kernel 3.0.7 Thank you very much. regards, tjiptomangun |
|||
| #12677 | obsolete | AHCI and ACPI seem to conflict | ||
| Description |
Hello, Starting from the launch of the 4.3.x branch there seem to be serious problems with the stability of the guests. In my case I get a lot of crashes with Windows 7 (32 bit) and Windows 2008R2 guests on my host; I tried using all the 3.11.x, 3.12.x and 3.13.x kernel versions on my host but the result is invariably the same - BSOD or sudden reset of the guest VM. I disabled 3D, 2D acceleration, tried IDE or SATA controllers, used 1 or 2 VCPUs, enabled or disabled the USB controller ... Unfortunately, the guests are simply not reliable any more! Going to older versions seems to stabilise the same guests - had used them for days without any sort of crash; going back to the 4.3.x branch will immediately display the behavior described. What intrigues me is the presence of the following lines in the logs: ... AHCI#0: Reset the HBA Reset initiated by ACPI ... I will upload some of the logs and also some of the dump files. Is this a known bug, a configuration-specific bug or a new one? Best regards, Marian |
|||
| #13474 | fixed | AHCI interrupt acknowledge violates specification => fixed in SVN | ||
| Description |
While optimizing FreeBSD AHCI driver I found that VirtualBox does not allow to acknowledge AHCI interrupt by writing AHCI IS (Interrupt Status) register without reading from that register first. You may see it looking inside HbaInterruptStatus_r() function, which is the only place where u32PortsInterrupted variable is cleared. But in configuration with only one SATA port present and command completion coalescing disabled I don't really need to read IS register to know the source of the interrupt. So in my driver I am just writing to it without spending extra time on read, which can be quite expensive. It perfectly works on real hardware, but not inside VirtualBox. I believe it is an AHCI specification violation from the VirtualBox side, and it should be fixed. |
|||

