On Thu, Feb 06, 2025 at 05:34:58PM -0500, Raymond Mao wrote: > Hi Simon, > > On Thu, 6 Feb 2025 at 10:41, Simon Glass <s...@chromium.org> wrote: > > > > Hi Raymond, > > > > On Wed, 5 Feb 2025 at 08:25, Raymond Mao <raymond....@linaro.org> wrote: > > > > > > +CC Ilias, > > > > > > Hi Simon, > > > > > > On Tue, 4 Feb 2025 at 20:57, Simon Glass <s...@chromium.org> wrote: > > > > > > > > The logic of this has become too confusing. > > > > > > > > The primary issue with the patch is that U-Boot needs to set up a > > > > bloblist in the first phase where BLOBLIST is enabled. Subsequent > > > > phases can then use that bloblist. > > > > > > > > But the first phase of U-Boot cannot assume that one exists. > > > > > > > > Reverting this commit seems like a better starting point for getting > > > > things working for all use-cases. > > > > > > > > This reverts commit 66131310d8ff1ba228f989b41bd8812f43be41c3. > > > > > > > > https://lore.kernel.org/u-boot/CAPnjgZ3hMHtiH=f5zkxnniofv_-vfryq1gn7qz5hku8wjo8...@mail.gmail.com/ > > > > Signed-off-by: Simon Glass <s...@chromium.org> > > > > --- > > > > > > > > > > If my understanding is correct, you want to add some logic to control > > > when the U-Boot should or should not get the bloblist from the > > > existing register argument. > > > But xferlist_from_boot_arg() should be called when a valid register > > > argument is there, I didn't see this in your patch. > > > Maybe you plan to do this with other patch series, but simply > > > reverting this results in a breaking of handoff policy and the > > > firmware handoff won't work. > > > > Yes, I certainly did not want to revert it, but the current code is > > too hard to understand and I did not look at it at the time it went > > in. I've had three tries at working with what you have here, but each > > turns to spaghetti. > > > > Still not very clear on what concerns you have and what is the way you > want to go. > The logic is straight forward, when U-Boot has a previous loader and > the registers pass in valid arguments - It indicates handoff should be > done using the transfer list. > Other kconfig options decide whether to use the passed in address > directly or copy to a predefined address. > But in either way, xferlist_from_boot_arg() is doing the right thing > to get the transfer list from the register if it exists and is valid. > I don't see a reason for removing it.
Exactly. We aren't reverting anything in master. As far as I can tell the only problems we have are: - No mechanism within U-Boot to pass a bloblist/handoff from one stage to another WITHOUT it being at a fixed address. - The actions of "is there a bloblist" and "make a bloblist" are tied together, which is a problem with a fixed address being set in memory that is either maybe uninitialized or could be re-initialized. But if we fix the former we don't really need to have the latter normally. -- Tom
signature.asc
Description: PGP signature
