Re: [go-nuts] Unexpected behaviour with LD_PRELOAD and fopen and cgo

Ah ok. So I managed to get something to compile (ended up having to cast the 'const char' arguments to fopen64 to char before passing them to my go_fopen64 function but I still get: fatal: morestack on g0 zsh: trace trap LD_PRELOAD=./preload python test.py foo I'm clearly out of my depth here

Re: [go-nuts] Unexpected behaviour with LD_PRELOAD and fopen and cgo

On Mon, Jan 15, 2018 at 12:36 PM, wrote: > Thanks Ian. I assume I'd include it like this: > // #include myfopen.h > And also a myfopen.c with the actual function > > But how do I ensure the fopen64 in myfopen.h is exported to override the > default? The fopen64 function written in myfopen.c will

Re: [go-nuts] Unexpected behaviour with LD_PRELOAD and fopen and cgo

Thanks Ian. I assume I'd include it like this: // #include myfopen.h And also a myfopen.c with the actual function But how do I ensure the fopen64 in myfopen.h is exported to override the default? On Monday, January 15, 2018 at 3:05:32 PM UTC-5, Ian Lance Taylor wrote: > > The simplist workarou

Re: [go-nuts] Unexpected behaviour with LD_PRELOAD and fopen and cgo

On Mon, Jan 15, 2018 at 11:54 AM, wrote: > > Thanks Ian. This did the trick and now I'm faced with the problem of > function signatures that include 'const'. On my system, fopen is declared > thus: > extern FILE *fopen (char *p0, char *modes) __wur > But fopen64 is: > extern FILE *fopen64 (const

Re: [go-nuts] Unexpected behaviour with LD_PRELOAD and fopen and cgo

Thanks Ian. This did the trick and now I'm faced with the problem of function signatures that include 'const'. On my system, fopen is declared thus: extern FILE *fopen (char *p0, char *modes) __wur But fopen64 is: extern FILE *fopen64 (const char *__restrict __filename, const char *__restrict __

Re: [go-nuts] Unexpected behaviour with LD_PRELOAD and fopen and cgo

On Mon, Jan 15, 2018 at 9:09 AM, wrote: > > I'm doing more experiments with created shared-object files with functions > that override syscalls and I've hit something I'd like to see if you guys > can help with. > > See a cut-down version of what I'm doing here > > There are 2 problems.. > > 1. I

[go-nuts] Unexpected behaviour with LD_PRELOAD and fopen and cgo

Hi there, I'm doing more experiments with created shared-object files with functions that override syscalls and I've hit something I'd like to see if you guys can help with. See a cut-down version of what I'm doing here There are 2 problems.. 1. If I r