RE: how Helix deal with the scenarion that ausbdisk is pluged out during playback?

　　I haven't fixed the issue, but got some clue:
　　1. memory map is used for file-io, see common/fileio/mmapmgr.cpp.
　　   In GetBlock(), the pBuffer is calculated from the page information, it may be invalid, then when there is a memcpy(), it will crash the engine.
　　On Windows system, it used ::IsBadReadPtr()to catch the exception as early as possible. But there is no such check (error handle) for Linux platform.
　　2. I tried to find some code works like IsBadReadPtr(), and wrap that in mmapmgr.cpp. see attachment. That works some to me, but need more debug.
　　3. after some try above, I found some other issue in smplfsys.cpp
　　In function CheckForCorruptFile(), I found it fail to determine it is  a corrupt file because m_ulSize equals to 0.  it is unexpected.
　　
　　BR
　　
　　
　　>-----Original Message-----
　　>From: [email protected]
　　>[mailto:[email protected]] On Behalf Of Zhao,
　　>Halley
　　>Sent: 2008年7月16日 8:40
　　>To: [email protected]; [email protected]
　　>Subject: RE: [Helix-client-dev] how Helix deal with the scenarion that
　　>ausbdisk is pluged out during playback?
　　>
　　>　　I met the same symptom as you.
　　>　　Smplfsys think it has got enough data without error.
　　>
　　>　　Where is the IHXDataFile for Linux system? Is it
　　>common/fileio/bufdataf.cpp?
　　>
　　>
　　>　　>-----Original Message-----
　　>　　>From: Eric Hyche [mailto:[email protected]]
　　>　　>Sent: 2008年7月15日 22:32
　　>　　>To: Zhao, Halley; [email protected]
　　>　　>Subject: RE: [Helix-client-dev] how Helix deal with the scenarion that
　　>a
　　>　　>usbdisk is pluged out during playback?
　　>　　>
　　>　　>
　　>　　>Halley,
　　>　　>
　　>　　>Extensive work was done in filesystem/local/full/smplfsys.cpp
　　>　　>to handle cases when the underlying IHXDataFile object
　　>　　>returns 0 bytes of returns less than requested (or returns
　　>　　>an error).
　　>　　>
　　>　　>However, in your case, it looks like the IHXDataFile
　　>　　>is saying it returned enough data, so the code to
　　>　　>handle these cases in smplfsys.cpp is not being
　　>　　>triggered.
　　>　　>
　　>　　>When the USB disk is unplugged, what is the behavior
　　>　　>of the IHXDataFile? Does it return 0 bytes when a
　　>　　>::Read() is requested? Does it return an error via
　　>　　>IHXDataFile::GetLastError()?
　　>　　>
　　>　　>Eric
　　>　　>
　　>　　>=============================================
　　>　　>Eric Hyche ([email protected])
　　>　　>Technical Lead
　　>　　>RealNetworks, Inc.
　　>　　>
　　>　　>> -----Original Message-----
　　>　　>> From: [email protected]
　　>　　>> [mailto:[email protected]] On
　　>　　>> Behalf Of Zhao, Halley
　　>　　>> Sent: Monday, July 14, 2008 5:45 AM
　　>　　>> To: [email protected]
　　>　　>> Subject: [Helix-client-dev] how Helix deal with the scenarion
　　>　　>> that a usbdisk is pluged out during playback?
　　>　　>>
　　>　　>> When Helix player is playing content on USB disk, then user
　　>　　>> unplug the USB disk, helix engine will crash in such scenario.
　　>　　>>
　　>　　>> I test it for helix-player and player_gtk_test on Linux,
　　>　　>> helix engine will crash in such scenario.
　　>　　>>
　　>　　>> On Windows system, Realplayer works well in such scenario,
　　>　　>> the presentation stopped, but you could open another stream.
　　>　　>>
　　>　　>>
　　>　　>>
　　>　　>> When I looked into some source code at
　　>　　>> filesystem/local/full/smplfsys.cpp,
　　>　　>>
　　>　　>> I found there seems to be some function to deal with such
　　>　　>> scenario: CheckForCorruptFile().
　　>　　>>
　　>　　>> However, when I try to debug it, I found the function isn't
　　>　　>> invoked in the scenario above.
　　>　　>>
　　>　　>>
　　>　　>>
　　>　　>> And I found after the last call to FinishDoRead(), the engine crash.
　　>　　>>
　　>　　>> But in this last call, for the condition sentence to
　　>　　>> CheckForCorruptFile():
　　>　　>>
　　>　　>> if (!pBuffer || actual < m_ulPendingReadCount)
　　>　　>>
　　>　　>>
　　>　　>>
　　>　　>> the pBuffer is valid and actual = 2048, and
　　>　　>> m_ulPendingReadCount = 2048.
　　>　　>>
　　>　　>>
　　>　　>>
　　>　　>> I think the scenario should be a common error handle for
　　>　　>> source plugin,
　　>　　>>
　　>　　>> Did anybody meet such issue before? Or maybe someone fix it
　　>　　>> in other branch?
　　>　　>>
　　>　　>>
　　>　　>>
　　>　　>> BR
　　>　　>>
　　>　　>>
　　>　　>>
　　>　　>>
　　>　　>>
　　>　　>>
　　>　　>>
　　>　　>> ZHAO, Halley (Aihua)
　　>　　>>
　　>　　>> Email: [email protected] <mailto:[email protected]>
　　>　　>>
　　>　　>> Tel: +86(21)61166476
　　>　　>>
　　>　　>> iNet: 8821-6476
　　>　　>>
　　>　　>> SSG/OTC/UMD
　　>　　>>
　　>　　>>
　　>　　>>
　　>　　>>
　　>
　　>
　　>_______________________________________________
　　>Helix-client-dev mailing list
　　>[email protected]
　　>http://lists.helixcommunity.org/mailman/listinfo/helix-client-dev

_______________________________________________
Helix-client-dev mailing list
[email protected]
http://lists.helixcommunity.org/mailman/listinfo/helix-client-dev

#include <setjmp.h>
#include <signal.h>
// globals

unsigned char g_ucCalledByTestRoutine = 0;
jmp_buf g_pBuf;
void __cdecl MemTestHandler ( int nSig)
{
     if ( g_ucCalledByTestRoutine ) longjmp ( g_pBuf, 1);
}
bool _IsBadReadPtr(const void* pv, unsigned long ulSize)
{
    char* pc;
    void(*pPrev) ( int sig);
    g_ucCalledByTestRoutine = 1;
    if(setjmp ( g_pBuf))
    {
         return true;
    }
    pPrev = signal (SIGSEGV, MemTestHandler);
    pc = (char*) malloc ( ulSize);
    memcpy ( pc, pv, ulSize);
    free(pc);
    g_ucCalledByTestRoutine = 0;
    signal ( SIGSEGV, pPrev);
    return false;
}