On Tue, Aug 08, 2006, Keith Bennett wrote: > > Kernel memory is a precious resource; it cannot be swapped out or used by > > anything else in the kernel once allocated, so IMHO even a 512 byte buffer > > shouldn't stick around more than we need it to. Kmalloc will generally > > return dead buffers from the size-512 slab cache without having to do an > > alloc anyway. > > Okay.
I should also mention that in my version of the driver there are sysfs entries which set the time, query device properties, etc. so the buffer is required throughout the lifespan of the driver. The memory doesn't just sit there unused. However, that version isn't likely to be pushed upstream and I agree that it is pointless keeping a buffer which is only used at initialisation time. The patch looks okay but if the READ_10 is to be caught then the "in_storage_mode" flag will need to be stored somewhere. The original version added a bit to the "flags" variable, but this makes the patch slightly more intrusive. Is USB_STOR_TRANSPORT_ERROR the correct thing to return from karma_init? Most of the other drivers seem to return 0 or 1. Let me know what you want to do about the READ_10 support. Keith. ------------------------------------------------------------------------- Using Tomcat but need to do more? Need to support web services, security? Get stuff done quickly with pre-integrated technology to make your job easier Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642 _______________________________________________ linux-karma-devel mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/linux-karma-devel
