Date: Mon, 3 Nov 2008 18:49:57 -0500 From: John Baldwin <jhb@freebsd.org> To: freebsd-hackers@freebsd.org Cc: Patrick =?iso-8859-1?q?Lamaizi=E8re?= <patfbsd@davenulle.org> Subject: Re: problem with bus_dmamap_load_uio Message-ID: <200811031849.57232.jhb@freebsd.org> In-Reply-To: <20081103215021.0412b7b2@baby-jane-lamaiziere-net.local> References: <20081103215021.0412b7b2@baby-jane-lamaiziere-net.local>
next in thread | previous in thread | raw e-mail | index | archive | help
On Monday 03 November 2008 03:50:21 pm Patrick Lamaizi=E8re wrote: > Hello, >=20 > (8-Current/i386) >=20 > I've got a problem with bus_dmamap_load_uio(9). I want to use it to > map an uio for DMA with the Geode security block (glxsb(4)). >=20 > It works, i can make milions of crypto operations with cryptotest. >=20 > But when (i guess, i'm not sure) the free memory becomes low, > bus_dmamap_load_uio() fails with errno =3D=3D EFBIG. >=20 > I can reproduce the problem by running "periodic daily", "dd > if=3D/dev/zero of=3Dfoo", ... >=20 > By sample i run into problems with Mem: 33M Active, 310M Inact, 82M > Wired, 60M Buf, 69M Free. >=20 > When it fails the uio is : > uio_segflg =3D UIO_SYSSPACE > uio_iovcnt =3D 1,=20 > totlen =3D 16384 (the total length for iov) > uio_resid =3D 16384 >=20 > (i've got some failure with size between 4000 and 16384) >=20 > I don't understand why bus_dmamap_load_uio() cannot load the uio. Also > when it fails, i copy the uio into a buffer and then a > bus_dmamap_load() of the buffer always works. >=20 > The dma tag is allocated with: >=20 > bus_dma_tag_create(bus_get_dma_tag(sc->sc_dev),/* parent */ > 16, 0, /* alignments, bounds */ > BUS_SPACE_MAXADDR_32BIT, /* lowaddr */ > BUS_SPACE_MAXADDR, /* highaddr */ > NULL, NULL, /* filter, filterarg */ > 16384, /* maxsize */ > 1, /* nsegments */ > 16384, /* maxsegsize */ > BUS_DMA_ALLOCNOW, /* flags */ > NULL, NULL, /* lockfunc, lockarg */ > &sc->sc_dmat); >=20 > The dma map is created just before the bus_dmamap_load_uio() and > destroyed later. >=20 > (source : http://user.lamaiziere.net/patrick/glxsb.c , see > glxsb_crypto_encdec() ). >=20 > I'm trying to understand and fix this problem for several days, so any > idea will be very cool... Your dma tag only allows a single scatter/gather entry (nsegments). What i= s=20 happening is that under memory pressure, the virtual buffer in userland is= =20 not built from physically contiguous pages, so you would need multiple s/g= =20 entries to describe the buffer. Hence EFBIG. If your hardware can handle= =20 multiple s/g entries, then just increase 'nsegments' and handle the multipl= e=20 segments in your callback routine. =2D-=20 John Baldwin
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200811031849.57232.jhb>