Error handling

From WikiPrizm
Jump to: navigation, search

This page has not been completed. Parts may be missing or reorganized before completed. Information is provided as-is and may have errors.

CASIOWIN handles a big variety of errors, in more or less user-friendly ways, from user-space and fully expected errors (like, for instance, a syntax error on a math expression) to errors due to faulty program code.

The OS sets up CPU parameters so that it can handle exceptions, both expected ones that happen by design (e.g. TLB miss) and others that would ideally never be triggered, like those due to illegal opcodes or illegal memory access. The latter ones cause System ERRORs.

User-level expected errors

Errors for which the OS is well prepared and are handled gracefully include syntax errors on math expressions and BASIC programs, user inputs out of range, calculation results out of range, wrong number of elements for math or BASIC functions, invalid data types on fields, etc. These are errors that normally result from mistakes on the user side and which are absolutely expected.

For those errors, a message containing a line of text in the form "<something> ERROR" is shown. This can be "Syntax ERROR", "Conversion ERROR", "Range ERROR", etc. Some error messages are more friendly and descriptive, like "Can't solve! / Adjust initial / value or bounds. / Then try again." or "Complex Number In List". EXIT, QUIT or AC/On can be used to break out of these message boxes, usually allowing for solving the situation that caused the error message.

User-level "less expected" errors

There are also a number of situations for which the OS is prepared, but which are not meant to happen on a normal basis. This includes file access errors, data corruption errors, invalid file path errors, and some kinds of errors which very rarely happen during proccess switching. Error messages for these errors include "Data ERROR", "Data is protected" or "Application ERROR" [the latter appeared to me once, when switching between a strip and a eActivity document --Gbl08ma (talk) 18:51, 3 December 2014 (EST) ]. Like with more usual errors, these messages can be closed with EXIT, QUIT or AC/On and, generally, normal operation is resumed.

Another example of a handled but not-so-expected problem is when the user formats the storage memory on the computer during a USB connection. In this case, "File System ERROR" is shown and certain operations will not be available until the user initializes the calculator using the Reset menu on the System app.

Some of these gracefully-handled errors are mentioned in the "Error Message Table" section of the Appendix of the Prizm's software manual.

System ERROR

When some problem is not handled by the OS at a higher level, an exception occurs at the CPU level. These exceptions are most usually triggered by custom add-ins with bugs, and, in normal situations, are handled by the OS. The OS handles them by stopping execution of the current process to some extent (but not completely: user timers will still run, for example), showing a pop-up on screen with title "System ERROR" and some information about the fault:

1417561924.698.syserrR.png

From this pop-up, one can press the EXIT key to reboot. Usually this works, but it's possible that the system got into such a messed state, that it will not work, and one must reboot the calculator by other means. The EXE key can be used to continue execution on certain kinds of exceptions, such as address exceptionsConfirmed, but this is generally not advised, because it will only make the problem worse.

In the case of system errors during execution of the child process, it is sometimes possible to press the MENU key and get into the Main Menu, open another add-in or built-in app (therefore ending the current child process, and switching to a new one), and the calculator will keep working as usual without the need for rebooting. However, this is not recommended and, depending on the cause of the error, may not work, causing further system errors or other kind of undefined behavior and, possibly, data loss. One very real possibility, is that file handles from the previous child process will not be closed when leaving a misbehaving app this way, and file operations may have strange behaviors, especially for files touched by the misbehaving program.

The pop-up appears to be a normal message box, probably opened with MsgBoxPush. In most cases, it's possible to see the last screen contents from the misbehaving application behind the message box. The first three lines are always the same. The fourth line tells the type of exception and may be "ADDRESS(R)", "ADDRESS(W)", "PROTECTED(R)", "PROTECTED(W)", "Interrupt", "Illegal Code Err" or "TLB ERROR". More information about each error can be found in the Exception Types section. The fifth and sixth lines don't always appear; when they do, they provide information about the state of program execution at the time of the exception.

Exception Types

The fourth line on the System Error dialog tells you what exception the OS handled. This gives you an idea of what is wrong with the code at the PC. The Transition address is the address of the exception handler called. VBR is the address stored at 0xA0000000.

ADDRESS(x)

Causes: This exception is thrown if there is a data address error. This can be caused if a word is accessed across word boundary, a longword across a longword boundary, or a quadword across a quadword boundary. It is also possible that this may be called when an instruction address is used that is not word-aligned. The (x) will either be R or W if the exception occurred on the Read or the Write cycle.

VBR Offset: 0x100

Exception Code: 0x0E0 on read, 0x100 on write

Priority: 5

Skippable: Yes

PROTECT(x)

Causes: This is unconfirmed, however this may be caused by a read or write to protected memory. The (x) will either be R or W if the exception occurred on the Read or the Write cycle.

VBR Offset: 0x100

Exception Code: 0x0A0 on read, 0x0C0 on write

Priority: 7

Skippable: Unknown

INTERRUPT

Causes: This is caused by either a TRAPA #imm (unconditional trap) assembly opcode or from an illegal slot instruction exception (See SH4a assembly manual for details).

VBR Offset: 0x100

Exception Code: 0x160

Priority: 4

Skippable: Unknown

An example of encountering this exception is by going to the diagnostic mode, pressing 3, 9, 2, F1. The "SYSTEM ERROR" message appears on screen with "INTERRUPT", but the target and PC lines are not shown. EXIT can't be used to reboot, despite what the text says.credits

Illegal Code Err

Causes: Assumed to be caused by either an unknown instruction being called or an opcode generated a general error (such as invalid usage of branch opcodes generating exceptions).

VBR Offset: 0x100

Exception Code: 0x180

Priority: 4

Skippable: Unknown

TLB ERROR

Causes: There was a memory access to the virtual memory that isn't mapped to a physical memory page in the UTLB page table (and possibly the ITLB page table if the access was from an instruction fetch) and the OS was unable to handle it. Note that the actual exception is normal behavior for the MMU and relies on the OS to handle it without generating errors. This occurs if the OS cannot handle it.

VBR Offset: 0x400

Exception Code: 0x040 if this was from a TLB miss in the UTLB cache from a read (Also a miss in the ITLB if this is an instruction fetch), 0x060 if this was from a TLB miss in the UTLB page table from a memory write.

Priority: 2 if from an ITLB miss, 6 if from a UTLB miss.

Skippable: Unknown

Bootloader integrity problems

The OS checks the bootloader integrity, or at least its presence, every time the calculator is turned on. If the bootloader is damaged, for example, by erasing the first flash sector, and no other damage is done, the calculator will keep operating until the OS attempts to reboot or batteries are taken off (or become empty). During the period the Prizm and its OS still work, there are no changes to normal operation except the following:

  • Every time the calculator is turned on, the diagnostic mode will open with the message "ABS Mark NG" on the top left, in the place of "LY755D MAIN":

1406915278.539.abs1.jpg

"ABS Mark NG" is shown whenever the diagnostic mode is entered and the 8 bytes starting at 0xA0000338 do not match "CASIOABS". If the only the "CASIOABS" string is damaged, the calculator will enter a emergency OS updating mode on the next reboot. This happens even in the "fx-CG10/20 Manager PLUS" emulator.

Performing a ROM checksum test from the diagnostic mode results in this:

1406915361.394.abs2.jpg

It is possible to close the diagnostic menu by pressing EXIT (unlike what happens when opened normally, it will not restart) and continue operating the calculator, which may be useful during an exam, for example (assuming you don't know enough nor care about what's happening, and can keep calm).

  • When connecting to a computer through USB, instead of the usual dialog asking the user to select a mode, a "Receiving..." screen, not usually accessible, appears. This is the color-version equivalent of the screen fx-9860G users see when connecting to a computer through USB. The computer, if a Windows machine configured to look for drivers, will look for CESG502 drivers, the same used for communication with a fx-9860G.
    Presumably, in this mode the calculator is listening for Protocol 7.00 commands, which, eventually along with its command 1 subtype 56 (Upload and Run), can be used to recover a damaged bootloader.
  • OS updates using the official update bundles might not work. The updater will send a binary, presumably through Upload and Run, that in normal conditions is used to retrieve information about the calculator and the current OS, as well as receive and flash the new OS when proceeding. The information the updater wants to retrieve is in the first flash sector which, along with the bootloader, may be damaged, meaning this initial step can fail. In that case (the only studied one, so far), an error message will be shown on the computer and the calculator will not reboot or show any other change on screen. The updater payload doesn't respond to keyboard interaction.
    Further executing the updater bundle will do nothing as it cannot find the calculator in the correct state for receiving the updater payload (it doesn't expect it to be running the payload already). The only way to exit from a binary sent through Upload and Run is rebooting the calculator, which likely means killing the calculator as the bootloader is damaged.