Re: initmap code

2004-09-09 Thread Alan Cox
On Iau, 2004-09-09 at 00:45, Ian Romanick wrote: So, what happens in x86-64 where sizeof(unsigned long) is 4 bytes for IA-32 apps and 8 bytes for x86-64 apps? I guess since it's a parameter to the ioctl (rather than embedded in a structure) it should be okay, but I just want to be sure...

Re: initmap code

2004-09-09 Thread Simon 'corecode' Schubert
On 09.09.2004, at 15:02, Alan Cox wrote: So, what happens in x86-64 where sizeof(unsigned long) is 4 bytes for IA-32 apps and 8 bytes for x86-64 apps? I guess since it's a parameter to the ioctl (rather than embedded in a structure) it should be okay, but I just want to be sure... Be cautious

initmap code

2004-09-08 Thread Simon 'corecode' Schubert
hey, please forgive my ignorance, as I'm not experienced with this matter. I'm in progress in porting the latest DRM changes to BSD. Jon's addmap-base-2 patch adds the function initmap(): int DRM(initmap)( drm_device_t *dev, unsigned int offset, unsigned int size, int type, int flags ) first

Re: initmap code

2004-09-08 Thread Jon Smirl
On Wed, 8 Sep 2004 12:38:54 +0200, Simon 'corecode' Schubert [EMAIL PROTECTED] wrote: hey, please forgive my ignorance, as I'm not experienced with this matter. int DRM(initmap)( drm_device_t *dev, unsigned int offset, unsigned int size, int type, int flags ) first of all, shouldn't

Re: initmap code

2004-09-08 Thread Jon Smirl
drm.h is included by user space programs (the mesa library). Is caddr_t a both a user space and kernel type or is it kernel only? Any types have to be both user and kernel space. On Thu, 9 Sep 2004 00:54:45 +0200, Simon 'corecode' Schubert [EMAIL PROTECTED] wrote: On 09.09.2004, at 00:24, Jon

Re: initmap code

2004-09-08 Thread Simon 'corecode' Schubert
On 09.09.2004, at 00:24, Jon Smirl wrote: and then, also concerning the declaration of drm_map_t: shouldn't be offset and size be of type caddr_t and size_t? (if this has been handled on list before I joined, please forgive me and tell me to RTFA) drm_map_t is compiled into X binaries so we can't

Re: initmap code

2004-09-08 Thread Adam Jackson
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 On Wednesday 08 September 2004 19:00, Jon Smirl wrote: drm.h is included by user space programs (the mesa library). Is caddr_t a both a user space and kernel type or is it kernel only? Any types have to be both user and kernel space. According to

Re: initmap code

2004-09-08 Thread Simon 'corecode' Schubert
On 09.09.2004, at 01:00, Jon Smirl wrote: drm.h is included by user space programs (the mesa library). Is caddr_t a both a user space and kernel type or is it kernel only? Any types have to be both user and kernel space. ok, sorry, pilot error. caddr_t is BSDism. so the proper way would be to

Re: initmap code

2004-09-08 Thread Ian Romanick
Simon 'corecode' Schubert wrote: caddr_t is a pointer to a kernel address. I'm not sure about current 64 bit implementations, but at least for ia32/linux/bsd sizeof(caddr_t) == sizeof(unsigned long). If this is the same on 64 bit architectures, a change would be no problem. It's just: