Hi Jan, On Tue, 7 Feb 2023 at 04:56, Jan Lübbe <jlu@xxxxxxxxxxxxxx> wrote: > > On Mon, 2023-02-06 at 17:32 -0600, Rob Herring wrote: > > +boot-architecture > > > > On Mon, Feb 6, 2023 at 3:25 PM Simon Glass <sjg@xxxxxxxxxxxx> wrote: > > > > > > Hi Rob, > > > > > > On Mon, 6 Feb 2023 at 10:15, Rob Herring <robh@xxxxxxxxxx> wrote: > > > > > > > > On Sat, Feb 4, 2023 at 6:04 AM Simon Glass <sjg@xxxxxxxxxxxx> wrote: > > > > > > > > > > Hi Peter, > > > > > > > > > > On Sat, 4 Feb 2023 at 02:36, Peter Robinson <pbrobinson@xxxxxxxxx> > > > > > wrote: > > > > > > > > > > > > Hi Simon, > > > > > > > > > > > > Does it make sense to devise something that is compatible with the > > > > > > kernel's pstore [1] mechanism? > > > > > > > > > > Possibly...can you please be a little more specific? > > > > > > > > Peter is talking about the same thing I suggested on IRC. > > > > > > > > pstore == ramoops > > > > > > Oh, I only looked at the DT binding as I thought that was what you > > > were talking about on irc. > > > > The binding is called ramoops as it's for the RAM backend for pstore. > > > > My suggestion was either using/extending ramoops or following its > > design as a reserved memory region. All you would need to extend the > > ramoops binding is a new property to define the size of your data. > > > > > For pstore, isn't the point that Linux wants to save stuff to allow > > > debugging or collection on reboot? What does that have to do with > > > console logs from firmware? That seems like a different thing. Or are > > > you suggesting that we add a pstore driver into U-Boot? It is quite a > > > lot of code, including compression, etc. It might be easier for Linux > > > to write the data into pstore when it starts up? > > > > Originally ramoops was just what you described. It has grown to > > multiple backends and types of records (hence the rename to pstore). > > If you just add a new subsection within the pstore region, then I > > think the existing kernel infrastructure will support reading it from > > userspace. Maybe new types have to be explicitly supported, IDK. > > > > U-boot being able to read pstore wouldn't be a terrible feature to > > have anyways if your boot crashes before anything else is up to get > > the output. Note I'd guess the ram backend doesn't do compression as > > supporting slightly corrupted ram is a feature which wouldn't work. > > This is basically how it works in Barebox. It can display the pstore contents > after a kernel crash and also (optionally) log to the pstore/ramooms console > log. Slight RAM corruption can be handled by using error correcting codes. > > It's not perfect, of course, but still very useful. Thanks for the pointer. I had a look at this. How do you deal with updating a filesystem that might be corrupt? Is that even a good idea, if the purpose of it is to collect data from a kernel crash? We are working on a firmware 'Transfer List' which is a simple data structure to communicate through the different firmware phases. Since U-Boot is the last one, in this case, I suppose it could do the ramoops thing and add files for each of the firmware phases. What about logging support? It would be nice to have a format that understands logging level, category, filename/function, etc. > > Regards, > Jan > > > I think any new DT binding is premature and pstore/ramoops was just a > > suggestion to consider. This needs wider consideration of how to > > handle all the various (boot) firmware logs. I've added the > > boot-architecture list for a bit more visibility. If this needs a call, I have not seen one for quite a while. It seems to get cancelled at the last minute. I would be happy to attend one to discuss this topic. But if people have ideas here, please weigh in. Regards, Simon
