Page 36 of 66 FirstFirst ... 26313233343536373839404146 ... LastLast
Results 701 to 720 of 1320
  1. #701  
    Installer is a Tapwave employee. I've already emailed him about this, but I wouldn't hold my breath. Obviously, it's not in Tapwave's best interest to help out the competition's customers.

    As for the beta 1 package...no luck. I get an error message stating that I'm not using a palmOne supported card (remember I've got a SanDisk v2.1 card).

    Scott
    Now THIS is the future of smartphones.
  2. #702  
    Quote Originally Posted by Scott R
    Installer is a Tapwave employee. I've already emailed him about this, but I wouldn't hold my breath. Obviously, it's not in Tapwave's best interest to help out the competition's customers.

    As for the beta 1 package...no luck. I get an error message stating that I'm not using a palmOne supported card (remember I've got a SanDisk v2.1 card).

    Scott
    Oops, forgot about the card type... Maybe later someone else can try it, or when I get home I will (boy, I think I've hard reset this poor 650 100 times already!)...
  3. #703  
    By the way, euroclie, I answered my own question about the creator id's...

    Types and creator IDs are case-sensitive, and are composed of four ASCII characters in the range 32-126 (decimal). Types and creator IDs consisting of all lowercase letters are reserved for use by Palm Inc., so any type or creator ID that you choose must contain at least one uppercase letter, digit, or symbol1.
  4. #704  
    Quote Originally Posted by shadowmite
    By the way, euroclie, I answered my own question about the creator id's...
    Glad to know you got the answer. You didn't yet answer to my question about modifying files which have an overlay (in which case the overlay would stop working). So did you modify any? (this would save me the hassle to have to hex-compare every file to find out which ones were modified... )
  5. #705  
    Euroclie, I just re-contructed all the files exactly as the installer would have done it. All the checksums should be accurate. I don't think that's the issue... I'm still wondering if we can do this a easier way. Can anyone read this:

    00000ca6 7600 MOVEQ #0,D3
    00000ca8 3f3c0011 MOVE.W #17!$11,-(A7)
    00000cac 2f3c6e657477 MOVE.L #1852142711!$6e657477,-(A7)
    00000cb2 2f3c44415441 MOVE.L #1145132097!$44415441,-(A7)
    00000cb8 4e4fa075 TRAP #15
    00000cbc DC.W sysTrapDmOpenDatabaseByTypeCreator

    And figure out what creator ID is being looked up?? I'm still a little new to palm API calls... The 4e4f is the systrap, and sysTrapDmOpenDatabaseByTypeCreator = 0x75. But what are those moves setting up? Is it hex? Octal!? Ack!
  6. #706  
    Quote Originally Posted by shadowmite
    No problem... Once again, this is the beta1 file. This is the closest we have come. No other releases changed from the results this will give you. What I suggest it just to install all files in this, but leave out any with Vppi or VPN in them... This may or may not help, it's a shot in the dark.
    My delay was due to trying several different file combinations.

    So I can confirm that unless you are trying to setup a wifi vpn (which now crashes) the VPN files weren't doing anything. Same packet trace.

    What creates the NetServLib you gave us? I'm noticing that it has calls to both CDMA and WiFi networking in it.

    Sorry,
    Adam
  7. #707  
    The debugger decodes the instruction opcodes and the data in both hex and decimal. Per below, $6e657477 is ascii bytes for 'netw' and 44415441 is 'DATA'.

    Quote Originally Posted by shadowmite
    Euroclie, I just re-contructed all the files exactly as the installer would have done it. All the checksums should be accurate. I don't think that's the issue... I'm still wondering if we can do this a easier way. Can anyone read this:

    00000ca6 7600 MOVEQ #0,D3
    00000ca8 3f3c0011 MOVE.W #17!$11,-(A7)
    00000cac 2f3c6e657477 MOVE.L #1852142711!$6e657477,-(A7)
    00000cb2 2f3c44415441 MOVE.L #1145132097!$44415441,-(A7)
    00000cb8 4e4fa075 TRAP #15
    00000cbc DC.W sysTrapDmOpenDatabaseByTypeCreator

    And figure out what creator ID is being looked up?? I'm still a little new to palm API calls... The 4e4f is the systrap, and sysTrapDmOpenDatabaseByTypeCreator = 0x75. But what are those moves setting up? Is it hex? Octal!? Ack!
  8. #708  
    Quote Originally Posted by shadowmite
    Euroclie, I just re-contructed all the files exactly as the installer would have done it. All the checksums should be accurate. I don't think that's the issue...
    OK, good, that point is settled at least...
    MOVE.L #1852142711!$6e657477,-(A7)
    MOVE.L #1145132097!$44415441,-(A7)
    Those "MOVE" instructions should load a Long integer value (1852142711 decimal, or 6E657477 in hex or "netw" in ascii) for the first move, and "DATA" for the second move, into the memory location refered to by -(A7)... That's what I understand after doing some googling about the MOVE.L a68k instruction, though, I don't understand it more than the average (beginner) C programmer would!
  9. #709  
    Hey Santa8Claws, good timing!
  10. #710  
    Lol, I'm still waiting for someone to point out the fact that there was a 9 in it and I suggested octal...

    Thanks for the decode, I've been busy tracing out the WifiGetSignalStrength routine... We might just be able to continue on this I guess...
  11. #711  
    Quote Originally Posted by AdamP
    My delay was due to trying several different file combinations.

    So I can confirm that unless you are trying to setup a wifi vpn (which now crashes) the VPN files weren't doing anything. Same packet trace.

    What creates the NetServLib you gave us? I'm noticing that it has calls to both CDMA and WiFi networking in it.

    Sorry,
    Adam
    The NetServLib came in the T5 driver bundle actually. I'm not certain if cdma might stand for something else in the network world... it's obviously code division multiple access, but could that be something other than cellular or did they compile in a header that they didn't need?

    Thanks for verifying the VPN files are not the culprit, if we didn't check, well you know murphy...

    I'm still tracing the way threw the wifipanel code. I'm not certain if we can change anything there. The nppi code has a call to the net.lib, but I'm guessing that accesses the default library on the phone. We need a way of hotplugging the fake net files in at the right time.
  12. #712  
    Quote Originally Posted by shadowmite
    The NetServLib came in the T5 driver bundle actually. I'm not certain if cdma might stand for something else in the network world... it's obviously code division multiple access, but could that be something other than cellular or did they compile in a header that they didn't need?
    FWIW, cdma is also the reverse of "amdc", which is the resource type for ARM assembler code.

    Have a look at any ARM application, you'll most of the time find the three "amdc", "amdd" and "amdi" resources. And remember that there's the little-endian vs. big endian problem which cause some 4-bytes integers (well, 2 bytes as well, of course) to be stored in the reverse order in some cases, and in the normal order in other places of the code.

    That's as much as I could tell by browsing the code with an hex editor, anyway!

    But you can have a look at this "hacking" thread on 1st Source or in Pilrc source code if you don't believe me!
  13. #713  
    thanks euroclie, I spaced that one...

    So, it looks like any library with the creator id of netl upon a reset takes over the system network library handling... But does it have to happen at a reset?
  14. #714  
    Quote Originally Posted by shadowmite
    Euroclie, I just re-contructed all the files exactly as the installer would have done it. All the checksums should be accurate. I don't think that's the issue... I'm still wondering if we can do this a easier way. Can anyone read this:

    00000ca6 7600 MOVEQ #0,D3
    00000ca8 3f3c0011 MOVE.W #17!$11,-(A7)
    00000cac 2f3c6e657477 MOVE.L #1852142711!$6e657477,-(A7)
    00000cb2 2f3c44415441 MOVE.L #1145132097!$44415441,-(A7)
    00000cb8 4e4fa075 TRAP #15
    00000cbc DC.W sysTrapDmOpenDatabaseByTypeCreator

    And figure out what creator ID is being looked up?? I'm still a little new to palm API calls... The 4e4f is the systrap, and sysTrapDmOpenDatabaseByTypeCreator = 0x75. But what are those moves setting up? Is it hex? Octal!? Ack!
    Not sure if you resolved thie already: this is just assembly's way of preparing the right parameters for the DC.W sysTrapDmOpenDatabaseByTypeCreator function, which expects (at least) three things on the memory stack (a7):
    (1) the data type (here DATA)
    (2)and the creator code NETW
    (3) and the number 17 (decimal, or 11 hex).

    one more thing:
    #1145132097!$44415441 , the left number represents the decimal representation of the the hex number, ie 1145132097decimal = 0x 44415441
  15. #715  
    Quote Originally Posted by shadowmite
    thanks euroclie, I spaced that one...

    So, it looks like any library with the creator id of netl upon a reset takes over the system network library handling... But does it have to happen at a reset?
    Well, technically speaking, I think that any file loaded in RAM and which happens to have the same CreatorID & type as a file in ROM takes precedence on the ROM file. That's how you can have OS updates without having to reflash the ROM every time.

    As for knowing when exactly this precedence takes place, I'd venture that it's probably a "feature" of the database manager that happens everytime you open a file...

    The PalmOS reference doc says (in the DmGetNextDatabaseByTypeCreator section):
    Because databases are scattered freely throughout memory space,
    they are not returned in any particular order—any database matching the specified type/creator criteria can be returned.Thus, if the value of the onlyLatestVers parameter is false, this function may return a database which is not the most recent version matching the specified type/creator pair. To obtain only the latest version of a database matching the search criteria, set the value of the onlyLatestVers parameter to true.

    When determining which is the latest version of the database, RAM databases are considered newer than ROM databases that have the same version number. Because of this, you can replace any ROMbased application with your own version of it. Also, a RAM database on card 1 is considered newer than a RAM database on card 0 if the version numbers are identical.
  16. #716  
    onpaws, thanks, just checking how many people can read assembly today! No, I was being and lazy and just asked here rather than working on it. I'm very scattered in thoughts today and jumping all over the place. Maybe I jumped the gun saying we were at a dead end.

    Looking here:

    000001d4 4e56ffdc L16 LINK A6,#-36 ; Net.lib
    000001d8 2f03 MOVE.L D3,-(A7)
    000001da 486efffe PEA -2(A6)
    000001de 41fa007c LEA L23,A0
    000001e2 4850 PEA (A0)
    000001e4 4e4fa0ba TRAP #15
    000001e8 DC.W sysTrapSysLibFind
    000001e8 3600 MOVE.W D0,D3
    000001ea 504f ADDQ.W #8,A7
    000001ec 6706 BEQ L17
    000001ee 7000 MOVEQ #0,D0
    000001f0 3003 MOVE.W D3,D0
    000001f2 6062 BRA L22

    It loads the net library reference for the wifi driver... this is in wifinppi. So, can we change this to call a different resource by creator ID? Hmm??? Source code would make this so much easier...
  17. #717  
    shadowmite, I suppose that you already did that (yeah, tonight's the night for stating the obvious, but then, what else can I do to try feeling a bit useful ), but if not, you might want to have a look at the "Palm OS companion2.pdf" from the PalmSource doc, the chapter 7 (Network) deals about netlib and internet libraries. Here's a brief quote:
    Two different Palm OSŪ libraries provide network services to applications:
    • The net library provides basic network services using TCP and UDP via a socket API. This library is discussed in the section Net Library.
    • The Internet library builds on the net library to provide a socket-like API to high-level Internet protocols such as HTTP. This library is discussed in the section Internet
    Library.
    and later:
    IMPORTANT: Applications cannot directly use the net library to make wireless connections. Use the Internet library for wireless connections.
    I'm starting to read that chapter, maybe there's useful stuff in there...
  18. #718  
    Futhermore, here is L23:

    0000025c 4e65742e6c6962 L23 DC.B 'Net.lib'

    So it is LoadEffectiveAddress into A0 with "Net.lib"? So is it looking for it by name I guess? Hmm... 7 chars...

    Furthermore, the call to SysLibFind params:

    Parameters
    nameP
    Pointer to the name of a loaded library.
    refNumP
    Pointer to a variable for returning the library reference number (on failure, this variable is undefined)


    Euroclie, I was reading that earlier but didn't get that far before being sidetracked by work... Ever heard of a FLM 6?
  19. #719  
    Quote Originally Posted by shadowmite
    So, it looks like any library with the creator id of netl upon a reset takes over the system network library handling... But does it have to happen at a reset?
    The doc I mentioned in my last post includes this piece of informations:
    Most built-in libraries (NetLib, serial, IR) are preloaded automatically when the system is reset.
    So I guess the answer to your question is: yes, this has to happen at reset...
  20. #720  
    So I guess the answer to your question is: yes, this has to happen at reset...
    Actually, I found a few other systraps like SysLibLoad

Posting Permissions