System freezes while using network in virtualbox

KoOS network development
  • I repeated the same for Qemu 0.12.2 and VirtualBox 4.3.0. r.89960 - the problem is not detected.

    Qemu
    -net nic,model=rtl8139

    VirtualBox
    PCnet-FAST III (Am79C973)

    Host System: Windows 7 Home Premium 64-bit SP1

    Kolibri r.4497
    Всем чмоки в этом проекте! Засуньте эти 11 лет себе в жопу!
  • I tried VirtualBox with PCnet-FAST III (Am79C973)
    In this case I was clicked actively about 5 minutes to got freese. r4497
    Last edited by Leency on Tue Jan 21, 2014 9:20 pm, edited 1 time in total.
    Из хаоса в космос
  • BUT. Clock dots are blinking 18:08 > 18 08 > 18:08
    Из хаоса в космос
  • Leency wrote:BUT. Clock dots are blinking 18:08 > 18 08 > 18:08
    It makes sense to try viewtopic.php?f=42&t=2034
    Всем чмоки в этом проекте! Засуньте эти 11 лет себе в жопу!
  • Please test build #4510
    "Any intelligent fool can make things bigger, more complex, and more violent. It takes a touch of genius -- and a lot of courage -- to move in the opposite direction." Albert Einstein
  • Sorry, not fixed rev 4511
    Attachments
    22-01-2014 18-28-15.png
    22-01-2014 18-28-15.png (62.8 KiB)
    Viewed 8768 times
    Из хаоса в космос
  • I'm unable to reproduce this freeze with version #4512.
    If somebody is still able to reproduce, please let me know.
    "Any intelligent fool can make things bigger, more complex, and more violent. It takes a touch of genius -- and a lot of courage -- to move in the opposite direction." Albert Einstein
  • Sorry, still reproduced #4512.
    Из хаоса в космос
  • With English distro too.
    Из хаоса в космос
  • It seems latest call before freeze is SOCKET_block.
    Latest message on debug board was: "SOCKET_block: suspending thread: 26"
    Then mouse and keyboard became unuseable, clock keeps ticking. After a while the screensaver appears.

    Code: Select all

    SOCKET_block:
    
            DEBUGF  DEBUG_NETWORK_VERBOSE, "SOCKET_block: %x\n", eax
    
            push    eax
    
            pushf
            cli
    
            ; Set the 'socket is blocked' flag
            or      [eax + SOCKET.state], SS_BLOCKED
    
            ; Suspend the thread
            push    edx
            mov     edx, [TASK_BASE]
            mov     [edx + TASKDATA.state], 1               ; Suspended
    
            ; Remember the thread ID so we can wake it up again
            mov     edx, [edx + TASKDATA.pid]
            DEBUGF  DEBUG_NETWORK_VERBOSE, "SOCKET_block: suspending thread: %u\n", edx
            mov     [eax + SOCKET.TID], edx
            pop     edx
            popf
    
            call    change_task
            pop     eax
    
            DEBUGF  DEBUG_NETWORK_VERBOSE, "SOCKET_block: continuing\n"
    
            ret
    
    O mighty guru, what have I done wrong this time?
    "Any intelligent fool can make things bigger, more complex, and more violent. It takes a touch of genius -- and a lot of courage -- to move in the opposite direction." Albert Einstein
  • I am not a guru, but... probably overwriting code area or data for one area overlaps another area.
    This is just idea.
    Всем чмоки в этом проекте! Засуньте эти 11 лет себе в жопу!
  • Then mouse and keyboard became unuseable, clock keeps ticking
    This usually means that some high-priority IRQ handler hangs somewhere, maybe blocked while waiting for something - which it shouldn't do in first place - or, less probably, does not return to the caller due to stack issues. When any IRQ is generated, IRQ controller blocks any other IRQ with same and lower priority, and the code must notify IRQ controller when IRQ handling is done; the code that calls a handler is responsible for that. Timer is IRQ0 and has the highest possible priority, so clock keeps ticking, scheduler keeps switching threads and so on. The priority of USB controller is variable, in particular, it can be lower than the priority of network IRQ, in which case USB would be blocked by broken network IRQ handler.
    Сделаем мир лучше!
  • There was a bug in SOCKET_notify which might have resulted in stack corruption in some cases.
    Please test #4528
    "Any intelligent fool can make things bigger, more complex, and more violent. It takes a touch of genius -- and a lot of courage -- to move in the opposite direction." Albert Einstein
  • hidnplayr wrote:Please test #4528
    VirtualBox 4.3.0. r.89960 - all 5 configurations are working properly.
    Всем чмоки в этом проекте! Засуньте эти 11 лет себе в жопу!
  • Who is online

    Users browsing this forum: No registered users and 5 guests