RPCS3/rpcs3
1
0
Fork 0
mirror of https://github.com/RPCS3/rpcs3.git synced 2025-12-06 07:12:28 +01:00
Code Issues Actions 1 Packages Projects Releases Wiki Activity

Updated Developer Information (markdown)

Alexandro Sánchez Bach 2014-05-24 03:20:36 -07:00
parent 4805227cab
commit 6437dd3b09
1 changed files with 13 additions and 10 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

23
Developer-Information.md

@ -28,8 +28,18 @@ Best wishes and happy coding!
## Information
#### Size of PS3 basic types
Checked with [PSL1GHT](http://psl1ght.com).
####OE bit
Some instructions contain an optional `OE`(overflow excpetion) flag. We currently do not implement those versions of the instructions. For example we emulate `add`, `addze`, `subfme` but not `addo`, `addzeo`, `subfmo` which are just the prior operations with the `OE` flag set.
This flag writes to the `OV` and `SO` (overflow and summary overflow) bits of the XER (Fixed-Point Exception Register) register. The reason why it's not implemented is that the [GCC code emitter](http://repo.or.cz/w/official-gcc.git/blob_plain/c2c80e70905bc2e5dedf8352588fc8cb10e3ec7d:/gcc/config/rs6000/rs6000.md) does not use these instructions and so we haven't found any PS3 software that uses it.
If we find some software that has some handwritten assembly that use those instructions we'll need to implement it (or just for completeness). A good reference for XER bits can be found in [PowerPC Microprocessor Family: The Programmer's Reference Guide ](https://www-01.ibm.com/chips/techlib/techlib.nsf/techdocs/852569B20050FF778525699600741775) chapter 1.3 and other linked references.
For the implementation of the interpreter it would probably be best to refrain from x86 assembly or compiler intrinsics like `__readeflags` to determine a overflow. Probably something from the [answers to this SO question](http://stackoverflow.com/questions/199333/best-way-to-detect-integer-overflow-in-c-c) is what we should consider.
####Misc
(Some of this information can be deleted, or at least properly organized).
* `sizeof(char)` = 1 (= 8 bits)
* `sizeof(short)` = 2
* `sizeof(int)` = 4
@ -38,11 +48,4 @@ Checked with [PSL1GHT](http://psl1ght.com).
* `sizeof(size_t)` = 4
* `sizeof(void*)` = 4
####OE bit
Some instructions contain an optional `OE`(overflow excpetion) flag. We currently do not implement those versions of the instructions. For example we emulate `add, addze, subfme` but not `addo, addzeo, subfmo` which are just the prior operations with the `OE` flag set.
This flag writes to the `OV` and `SO` (overflow and summary overflow) bits of the XER (Fixed-Point Exception Register) register. The reason why it's not implemented is that the [GCC code emitter](http://repo.or.cz/w/official-gcc.git/blob_plain/c2c80e70905bc2e5dedf8352588fc8cb10e3ec7d:/gcc/config/rs6000/rs6000.md) does not use these instructions and so we haven't found any PS3 software that uses it.
If we find some software that has some handwritten assembly that use those instructions we'll need to implement it (or just for completeness). A good reference for XER bits can be found in [PowerPC Microprocessor Family: The Programmer's Reference Guide ](https://www-01.ibm.com/chips/techlib/techlib.nsf/techdocs/852569B20050FF778525699600741775) chapter 1.3 and other linked references.
For the implementation of the interpreter it would probably be best to refrain from x86 assembly or compiler intrinsics like `__readeflags` to determine a overflow. Probably something from the [answers to this SO question](http://stackoverflow.com/questions/199333/best-way-to-detect-integer-overflow-in-c-c) is what we should consider
Reading from invalid addresses / Trying to execute code from invalid addresses (e.g. stack) / Trying to write to invalid addresses (e.g. functions) makes the application exit on real PS3s.