Messages in this thread | | | Subject | Re: Binary drivers | From | Timothy Writer <> | Date | 05 Dec 1999 22:05:58 -0500 |
| |
Bill Huey <billh@burn.ucsd.edu> writes:
> > I don't think so. Implementing a binary compatible API usually requires a > > level of indirection that makes optimizations like inlining impossible. > > [here it goes, with a possible foot in the mouth ;-)] > > Uh, how ?
To preserve binary compatibility, you have to account for things like structure layout and size changes. One way to do that is to hide structures behind a pointer, IOW replace structures with a pointer to an opaque structure. Access to such strucures is then provided through functions which can't be inlined without exposing the layout, precisely what you were trying to avoid.
> The header files needs to include for module compilation is identical > for both open source and binary only modules.
Says who?
> Plus code changes that affect modules seems pretty minimal across kernel > revisions. I'm not an authority about this. > > > > Solaris, Netware, QNX, BeOS, MacOS or any other commercial OS. Fact > > > is they all use binary device drivers, and many of them are a lot > > > more stable than Linux is. > > > .... and a _lot_ slower on the same hardware. > > Well, they're alot slower only in different circumstances for reason not > relevant to binary-only module support.
How do you know? Have you seen the source code? Isn't it possible that the ABI is a contributing factor (to lacklustre performace)? Okay, that still doesn't mean that it's impossible to design an ABI that doesn't have a performance impact. So, show us your design.
-- tim writer <tim@starnix.com> starnix inc. tollfree: 1-87-pro-linux brampton, ontario, canada http://www.starnix.com professional linux services & products
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
| |