From: Pete Wyckoff <pw@xxxxxxx> Subject: [PATCH v2] bsg: iovec support with compat Date: Mon, 19 Mar 2007 14:07:27 -0400 > fujita.tomonori@xxxxxxxxxxxxx wrote on Mon, 19 Mar 2007 23:21 +0900: > > From: Douglas Gilbert <dougg@xxxxxxxxxx> > > Subject: Re: [PATCH] bsg: iovec support > > Date: Mon, 19 Mar 2007 10:04:45 -0400 > > > > > >> Pete is also suggesting (shown above) a revised sg_io_vec > > > >> structure that uses a uint64_t for the pointer to simplify > > > >> 32, 64 bit thunking. > > > > > > > > All I said is that it would be better to use the existing compat > > > > infrastructure (sg_build_iovec, sg_ioctl_trans, etc in > > > > fs/compat_ioctl.c) instead of adding another compat code. > > > > > > Won't sg v4 make this even a bigger mess, at least > > > initially anyway? > > > > Inventing a new iovec structure like sg_io_v4_vec and something like > > blk_rq_map_user_iov_sgv4 sounds a mess. > > It is sort of a mess to have new blk mapping routine for a new iovec > type. But I very much did not want to introduce the need for > another compat conversion. But, I took a look at how it would be to > pass the existing sg_iovec into bsg. > > Adding a new bsg_write_compat would be bad. There is lots of > parsing and setup before we even get to the possibility of iovec > data mapping. Reusing just sg_build_ioctl from compat_ioctl.c is > also suboptimal as this function is built around the idea of a > contiguous sg_io_hdr + iovec in userspace. The function is small > enough that splitting it into a generic + ioctl-specific part would > add too much abstraction to be worth it. > > Here is the patch to use sg_iovec, with its userspace void * and > size_t, and the CONFIG_COMPAT code to fixup 32-bit userspace. I'm > not fond of having __u64 for non-iovec buffer representations, and > void * for iovec buffer representations, but it saves having to > build an sg_iovec just to call into the existing blk_rq_map_user_iov. > > Comments? The compat code should not go to bsg.c. - To unsubscribe from this list: send the line "unsubscribe linux-scsi" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
