On Sun, Nov 16, 2025 at 10:28:28PM +0100, Marek Vasut wrote: > On 11/16/25 3:13 PM, Tom Rini wrote: > > On Sun, Nov 16, 2025 at 01:16:53AM +0100, Marek Vasut wrote: > > > On 11/14/25 3:36 PM, Tom Rini wrote: > > > > > > Hello Tom, > > > > > > > > We would need a wrapper for almost every libfdt call, which isn't > > > > > really > > > > > better. > > > > > > > > > > > What does fdt_getprop() return when the offset is -1 ? > > > > > If can_assume(VALID_DTB) , then crash on NULL pointer dereference . > > > > > If not, I think FDT_END (premature end) or an error code. > > > > > > > > How are we now getting in this state is a question. Are we being passed > > > > an unaligned DT and trying to work around/with it? > > > This is superseded by: > > > > > > [PATCH] xtensa: Assure end of U-Boot is at 8-byte aligned offset > > > > > > It turns out the problem was yet another unaligned appended DT, on xtensa > > > this time. > > > > I wonder if we have still more places. I've seen patches in your series > > for many but not all of > > https://lore.kernel.org/all/[email protected]/ > > and I don't recall how many of those were done individually instead > > afterwards. > Was there any reason why that patch (or subset of) was not picked up ?
It didn't get split up, and truncate isn't portable but later in the thread we confirmed how to use dd in a way that is and does what's required. -- Tom
signature.asc
Description: PGP signature
