On Tue, 19 May 2020 at 16:29, Russell King - ARM Linux admin <linux@xxxxxxxxxxxxxxx> wrote: > > On Tue, May 19, 2020 at 03:56:59PM +0200, Ard Biesheuvel wrote: > > On Tue, 19 May 2020 at 13:21, Geert Uytterhoeven <geert@xxxxxxxxxxxxxx> wrote: > > > > > > Hi Russell, > > > > > > CC devicetree > > > > > > On Tue, May 19, 2020 at 11:46 AM Russell King - ARM Linux admin > > > <linux@xxxxxxxxxxxxxxx> wrote: > > > > On Tue, May 19, 2020 at 11:44:17AM +0200, Geert Uytterhoeven wrote: > > > > > On Tue, May 19, 2020 at 10:54 AM Lukasz Stelmach <l.stelmach@xxxxxxxxxxx> wrote: > > > > > > It was <2020-04-29 śro 10:21>, when Geert Uytterhoeven wrote: > > > > > > > Currently, the start address of physical memory is obtained by masking > > > > > > > the program counter with a fixed mask of 0xf8000000. This mask value > > > > > > > was chosen as a balance between the requirements of different platforms. > > > > > > > However, this does require that the start address of physical memory is > > > > > > > a multiple of 128 MiB, precluding booting Linux on platforms where this > > > > > > > requirement is not fulfilled. > > > > > > > > > > > > > > Fix this limitation by obtaining the start address from the DTB instead, > > > > > > > if available (either explicitly passed, or appended to the kernel). > > > > > > > Fall back to the traditional method when needed. > > > > > > > > > > > > > > This allows to boot Linux on r7s9210/rza2mevb using the 64 MiB of SDRAM > > > > > > > on the RZA2MEVB sub board, which is located at 0x0C000000 (CS3 space), > > > > > > > i.e. not at a multiple of 128 MiB. > > > > > > > > > > > > > > Suggested-by: Nicolas Pitre <nico@xxxxxxxxxxx> > > > > > > > Signed-off-by: Geert Uytterhoeven <geert+renesas@xxxxxxxxx> > > > > > > > Reviewed-by: Nicolas Pitre <nico@xxxxxxxxxxx> > > > > > > > Reviewed-by: Ard Biesheuvel <ardb@xxxxxxxxxx> > > > > > > > Tested-by: Marek Szyprowski <m.szyprowski@xxxxxxxxxxx> > > > > > > > Tested-by: Dmitry Osipenko <digetx@xxxxxxxxx> > > > > > > > --- > > > > > > > > > > > > [...] > > > > > > > > > > > > Apparently reading physical memory layout from DTB breaks crashdump > > > > > > kernels. A crashdump kernel is loaded into a region of memory, that is > > > > > > reserved in the original (i.e. to be crashed) kernel. The reserved > > > > > > region is large enough for the crashdump kernel to run completely inside > > > > > > it and don't modify anything outside it, just read and dump the remains > > > > > > of the crashed kernel. Using the information from DTB makes the > > > > > > decompressor place the kernel outside of the dedicated region. > > > > > > > > > > > > The log below shows that a zImage and DTB are loaded at 0x18eb8000 and > > > > > > 0x193f6000 (physical). The kernel is expected to run at 0x18008000, but > > > > > > it is decompressed to 0x00008000 (see r4 reported before jumping from > > > > > > within __enter_kernel). If I were to suggest something, there need to be > > > > > > one more bit of information passed in the DTB telling the decompressor > > > > > > to use the old masking technique to determain kernel address. It would > > > > > > be set in the DTB loaded along with the crashdump kernel. > > > > > > > > > > Shouldn't the DTB passed to the crashkernel describe which region of > > > > > memory is to be used instead? > > > > > > > > Definitely not. The crashkernel needs to know where the RAM in the > > > > machine is, so that it can create a coredump of the crashed kernel. > > > > > > So the DTB should describe both ;-) > > > > > > > > Describing "to use the old masking technique" sounds a bit hackish to me. > > > > > I guess it cannot just restrict the /memory node to the reserved region, > > > > > as the crashkernel needs to be able to dump the remains of the crashed > > > > > kernel, which lie outside this region. > > > > > > > > Correct. > > > > > > > > > However, something under /chosen should work. > > > > > > > > Yet another sticky plaster... > > > > > > IMHO the old masking technique is the hacky solution covered by > > > plasters. > > > > > > > I think debating which solution is the hacky one will not get us anywhere. > > > > The simple reality is that the existing solution works fine for > > existing platforms, and so any changes in the logic will have to be > > opt-in in one way or the other. > > > > Since U-boot supports EFI boot these days, one potential option is to > > rely on that. I have some changes implementing this that go on top of > > this patch, but they don't actually rely on it - it was just to > > prevent lexical conflicts. > > > > The only remaining options imo are a kernel command line option, or a > > DT property that tells the decompressor to look at the memory nodes. > > But using the DT memory nodes on all platforms like this patch does is > > obviously just too risky. > > > > On another note, I do think the usable-memory-region property should > > be implemented for ARM as well - relying on this rounding to ensure > > that the decompressor does the right thing is too fragile. > > What is "too fragile" is trying to change this and expecting everything > to continue working as it did before. > That is my point. > How will switching to EFI help? Won't the crashdump kernel detect EFI > and try to get the memory map from EFI, whereby it runs into exactly > the same issue that the DT approach does? > No. If you boot from kexec, then the EFI stub is completely circumvented, and things work as before. > The current crashkernel situation works precisely because of the 128M > masking that is being done. > Indeed. That is precisely my point.