From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Date: Sun, 16 Dec 2007 22:35:02 +1100 From: David Gibson To: Milton Miller Subject: Re: [PATCH 1/7] bootwrapper: Add a firmware-independent "raw" target. Message-ID: <20071216113502.GB26307@localhost.localdomain> References: <20071213234221.B0BC27D0054@mail142-sin.bigfish.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii In-Reply-To: Cc: ppcdev List-Id: Linux on PowerPC Developers Mail List List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , On Fri, Dec 14, 2007 at 02:05:56AM -0600, Milton Miller wrote: > On Fri Dec 14 10:43:27 EST 2007, Stephen Neuendorffer wrote: > > > From: Grant Likely > > > > This target produces a flat binary rather than an ELF file, > > fixes the entry point at the beginning of the image, and takes > > a complete device tree with no fixups needed. > > > > The device tree must have labels on /#address-cells, the timebase > > frequency, and the memory size. > > > > Signed-off-by: Grant Likely > > --- > > > You indicated in the intro in 0/ that this was not ready, and you > didn't include your own s-o-b, but you did not put any statements to > that effect in the header. The intro is not copied into patchwork, > which maintainers often use when deciding what to push. > > Now on to why this should not be merged: > > In addition to the above, it changes the build rules. It tries to > change wrapper to assemble the .dtb into a .o from a .S file, but > doesn't set any flags to force the assembler into the right mode. In > contrast the linker is controlled by the .lds linker script. > > In addition, the requirement for assembly labels can easily be > eliminated. As mentioned above, they are used for 3 properties. With > the existing library (in 2.6.24 and earlier), call simple_malloc_init > with a small bss array (like BSS_STACK does to allocate stack), and > then read the properties out of the device tree. At that point, call > simple_malloc_init a second time using the found memory size. As I > said the last time this was posted, my patches to boot from kexec > implemented this strategy. > > However, with the new libfdt, which is already in for-2.6.25, we should > no longer need malloc() to simple read the tree. At least that is > what was advertised. That's correct. No malloc() is necessary for read-only access. -- David Gibson | I'll have my music baroque, and my code david AT gibson.dropbear.id.au | minimalist, thank you. NOT _the_ _other_ | _way_ _around_! http://www.ozlabs.org/~dgibson