-
Notifications
You must be signed in to change notification settings - Fork 79
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
bugfixes an mips start #63
Conversation
fixed missing native libs in jar
TODO move that in a global var???
better error message for test
…so fixed precsion .# Bitte geben Sie eine Commit-Beschreibung für Ihre Änderungen ein. Zeilen,
Gcc on i686(32bit) uses 96 bit for storage and 80 for long double itself.
…0 - as I would expect so there may is something somewhere.
Yes,
I know this.
Some tests on arm fail too.
I was diving in the c sources, and will investigate it later, If I can
debug that 10+ stacked defines.
|
… the size of the return value only on big endian. removed define to enable check and fix on little endian too.
@aploese Sorry for the long delay, I did not see this until now! Are we still good to go forward with these changes? Not much has happened in jffi since. |
Hi Charles,
after digging into jffi,jffi-jnr,jnr-posix I thougth there must be a
simpler solution.
See this here https://github.com/aploese/jnhw.
Why?
In short:
No jffi, just direct calls to the function. There must be a direct jni
mapping for each function. The Defines are pulled in by jni calls to.
The compiling is done with autotools - so there should be not too much
trouble compiling on each platform. Try it out!
From a security point of view it should be nearly impossible to produce
a SIGSEV which terminates the whole JVM - in I.M.H.O this happend too
often with jffi.
In jffi/jffi-jnr some bugs are fixed, but some may still exist in the
native assembler generation and in the bytecode assembler generation.
Currently JNHW is my tool of choice ;-).
Am Donnerstag, den 11.04.2019, 12:19 -0700 schrieb Charles Oliver
Nutter:
… @aploese Sorry for the long delay, I did not see this until now!
Are we still good to go forward with these changes? Not much has
happened in jffi since.
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub, or mute the thread.
|
So...it's autogenerated JNI? And you have to run a native build for each platform and library you intend to use, yes? The primary purpose of JNR, like FFI, is not having to rebuild anything for each platform and library you use. I guess that is not a requirement for you. JNHW would not work for any of the reasons I use JNR, unfortunately. Thanks for your effort! Perhaps someone else with MIPS expertise will pick it up so we can have MIPS support in the future. |
Am Donnerstag, den 11.04.2019, 14:05 -0700 schrieb Charles Oliver
Nutter:
> No jffi, just direct calls to the function. There must be a direct
> jni
> mapping for each function.
So...it's autogenerated JNI? And you have to run a native build for
each platform and library you intend to use, yes?
No its direct and handcrafted. after that a ./configure && make is all
you need to know of.
The primary purpose of JNR, like FFI, is not having to rebuild
anything for each platform and library you use.
You will have to rebuild jffi for each platform you intend to run.
I think the greatest risk of using jffi-jnr is to call a vararg
function with a floating point argument and ignoring to call it right
way.
If you are that deep in the native c-code, you should be able to write
the jni wrapper code as well - otherwise it would be better to leave it
alone...
I guess that is not a requirement for you. JNHW would not work for
any of the reasons I use JNR, unfortunately.
Thanks for your effort! Perhaps someone else with MIPS expertise will
pick it up so we can have MIPS support in the future.
—
Be aware that I fixed some bugs in my fork that I found and only
partially reported.
If I recall correctly some datatypes are not correct mapped in jffi-
jnr...
… You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub, or mute the thread.
|
This fixes #60, #61, #51
All native libs have to be recompiled!
mips test are passing