It does not have to do anything with USB. 
 
It happens an my Palm Vx (Serial port!) as well. 
 
BTW: kpilot has never ever synced my palm, while 
pilot-xfer -sync always works out of the box. 
 
The version I use is self-compiled CVS version: 
V4.3.5 (Using KDE 3.1.90 (CVS >= 20021124)) 
 
And here's the log: 
 
HotSync Log 
23:19:05 Trying to open device... 
23:19:05 Device link ready. 
23:19:15 Checking last PC... 
23:19:15 Syncing with KMail(Kroupware) disabled 
23:19:15 KPilot 4.3.5 HotSync starting...  
23:19:16 [Conduit vcal-conduit] 
23:19:17 Syncing with file "" 
23:20:52 [Conduit knotes-conduit] 
23:20:52 KNotes is not running. The conduit must be able to make a DCOP connection to 
KNotes for synchronization to take place. Please start KNotes and try again. 
23:20:55 The conduit knotes-conduit could not be executed. 
23:20:56 [Conduit abbrowser_conduit] 
23:22:25 No Files to install 
23:22:25 End of HotSync  
23:22:25 HotSync Completed. 
 
23:22:27 Pilot device /dev/pilot doesn't exist. Assuming the device uses DevFS. 
23:22:28 Trying to open device... 
23:22:28 Could not open device: /dev/pilot (will retry) 
 
 

Forget my comment: Error Message on Palm is the same (Invalid uniqueID passed),
but it happens while syncing adresses. Doing full sync combined with some other
"correct" syncing options helps.
Chris

I have the same problem when syncing  my Palm III, except the line number is
different.

I did have an address/contact conflict i.e. address was different on the PC and
Handheld, I got the conflict resolution dialog box with options to choose which
address should be used in sync'ing.  I selected "leave untouched", after that it
hung and I had to reset the Palm.  I retried and selected "PC overrides
Handheld" and got the same thing.

This is the error message:

DataMgr.c Line:4250
Invalid uniqueID passed

<reset>

Other info that maybe useful:
=======================
-- pilot-link.0.11.7
-- KPIM built from sources KDE3.1.0

g++ -v
Reading specs from /usr/lib/gcc-lib/i386-slackware-linux/2.95.3/specs
gcc version 2.95.3 20010315 (release)

-- Kevin

BTW, KPIM was built from sources KDE3.1.0/QT3.1.2

-- MG

Subject: kdepim/kpilot/lib

CVS commit by kainhofe: 

Added a check for invalid record ids when searching or writing records to the handheld. If they take more than 3 bytes, the handheld will crash, so I reset the id to zero to cause new record IDs to be assigned. This will prevent these crashes in the future, but I still have to find the place where these wrong values were assigned first place.

CCMAIL: 59313@bugs.kde.org
CCMAIL: 51216@bugs.kde.org


  M +3 -3      pilotRecord.h   1.7
  M +19 -3     pilotSerialDatabase.cc   1.16


--- kdepim/kpilot/lib/pilotSerialDatabase.cc  #1.15:1.16
@@ -144,4 +144,10 @@ PilotRecord *PilotSerialDatabase::readRe
                 return 0L;
         }
+        if (id>0xFFFFFF)
+        {
+                kdError() << k_funcinfo <<  " Encountered an invalid record id "
+                        <<id<<endl;;
+                return 0L;
+        }
         if (dlp_ReadRecordById(fDBSocket, getDBHandle(), id, buffer, &index,
                         &size, &attr, &category) >= 0)
@@ -210,5 +216,5 @@ PilotRecord *PilotSerialDatabase::readNe
 }
 
-// Writes a new record to database (if 'id' == 0, one will be assigned and returned in 'newid')
+// Writes a new record to database (if 'id' == 0 or id>0xFFFFFF, one will be assigned and returned in 'newid')
 recordid_t PilotSerialDatabase::writeRecord(PilotRecord * newRecord)
 {
@@ -222,4 +228,14 @@ recordid_t PilotSerialDatabase::writeRec
                 return 0;
         }
+        // Do some sanity checking to prevent invalid UniqueIDs from being written
+        // to the handheld (RecordIDs are only 3 bytes!!!). Under normal conditions
+        // this check should never yield true, so write out an error to indicate
+        // someone messed up full time...
+        if (newRecord->getID()>0xFFFFFF)
+        {
+                kdError() << k_funcinfo << "Encountered an invalid record id "
+                        <<newRecord->getID()<<", resetting it to zero.";
+                newRecord->setID(0);
+        }
         success =
                 dlp_WriteRecord(fDBSocket, getDBHandle(),
@@ -227,5 +243,5 @@ recordid_t PilotSerialDatabase::writeRec
                 newRecord->getCat(), newRecord->getData(),
                 newRecord->getLen(), &newid);
-        if (newRecord->getID() == 0)
+        if (newRecord->getID() != newid)
                 newRecord->setID(newid);
         return newid;

--- kdepim/kpilot/lib/pilotRecord.h  #1.6:1.7
@@ -51,6 +51,6 @@ public:
         void  setCat(int cat) { fCat = cat; }
 
-        unsigned long getID() const { return fID; }
-        void setID(unsigned long id) { fID = id; }
+        recordid_t getID() const { return fID; }
+        void setID(recordid_t id) { fID = id; }
 
 private:
@@ -59,5 +59,5 @@ private:
         int   fAttrib;
         int   fCat;
-        unsigned long fID;
+        recordid_t fID;
 
 public:

*** Bug 59313 has been marked as a duplicate of this bug. ***

The check for invalid IDs which I commited to current cvs and the 3.1 branch 
disallows the situation, where the crash occurs, and resets the ID to a valid value. So 
I consider this bug fixed. 

*** Bug 63785 has been marked as a duplicate of this bug. ***