| Message ID | 20230322170448.3600124-1-gsomlo@gmail.com |
|---|---|
| State | Superseded |
| Headers | show |
| Series | [v1] docs: Correct FW_JUMP_FDT_ADDR calculation example | expand |
On Wed, Mar 22, 2023 at 01:04:48PM -0400, Gabriel Somlo wrote: > When using `PLATFORM=generic` defaults, the kernel is loaded at > `FW_JUMP_ADDR = FW_TEXT_START + 0x200000`, and the FDT is loaded at > `FW_JUMP_FDT_ADDR = FW_TEXT_START + 0x2200000`. > > Therefore, the maximum kernel size before `FW_JUMP_FDT_ADDR` must > be increased is the difference, or `0x2000000`. Sounds good, but this is only correct for RV64. For RV32 it should be 0x1e00000. I wonder if we should do something like ``` ${CROSS_COMPILE}objdump -h $KERNEL_ELF | sort -k 5,5 | awk -n '/^ +[0-9]+ /\ {addr="0x"$3; size="0x"$5; printf "0x""%x\n",addr+size}' \ - | (( `tail -1` > 0x2200000 )) && echo fdt overlaps kernel,\ - increase FW_JUMP_FDT_ADDR + | (( `tail -1` > (FW_JUMP_FDT_ADDR - FW_JUMP_ADDR) )) && echo fdt overlaps \ + kernel, increase FW_JUMP_FDT_ADDR ${LLVM}objdump -h --show-lma $KERNEL_ELF | sort -k 5,5 | \ awk -n '/^ +[0-9]+ / {addr="0x"$3; size="0x"$5; printf "0x""%x\n",addr+size}'\ - | (( `tail -1` > 0x2200000 )) && echo fdt overlaps kernel,\ - increase FW_JUMP_FDT_ADDR + | (( `tail -1` > (FW_JUMP_FDT_ADDR - FW_JUMP_ADDR) )) && echo fdt overlaps \ + kernel, increase FW_JUMP_FDT_ADDR ``` instead? Thanks, drew > > Update the example calculation to reflect this, and avoid confusion. > > Signed-off-by: Gabriel Somlo <gsomlo@gmail.com> > --- > docs/firmware/fw_jump.md | 4 ++-- > 1 file changed, 2 insertions(+), 2 deletions(-) > > diff --git a/docs/firmware/fw_jump.md b/docs/firmware/fw_jump.md > index 956897e..d1b07b9 100644 > --- a/docs/firmware/fw_jump.md > +++ b/docs/firmware/fw_jump.md > @@ -48,12 +48,12 @@ follows: > ``` > ${CROSS_COMPILE}objdump -h $KERNEL_ELF | sort -k 5,5 | awk -n '/^ +[0-9]+ /\ > {addr="0x"$3; size="0x"$5; printf "0x""%x\n",addr+size}' \ > - | (( `tail -1` > 0x2200000 )) && echo fdt overlaps kernel,\ > + | (( `tail -1` > 0x2000000 )) && echo fdt overlaps kernel,\ > increase FW_JUMP_FDT_ADDR > > ${LLVM}objdump -h --show-lma $KERNEL_ELF | sort -k 5,5 | \ > awk -n '/^ +[0-9]+ / {addr="0x"$3; size="0x"$5; printf "0x""%x\n",addr+size}'\ > - | (( `tail -1` > 0x2200000 )) && echo fdt overlaps kernel,\ > + | (( `tail -1` > 0x2000000 )) && echo fdt overlaps kernel,\ > increase FW_JUMP_FDT_ADDR > ``` > > -- > 2.39.2 > > > -- > opensbi mailing list > opensbi@lists.infradead.org > http://lists.infradead.org/mailman/listinfo/opensbi
On Thu, Mar 23, 2023 at 06:18:00PM +0100, Andrew Jones wrote: > On Wed, Mar 22, 2023 at 01:04:48PM -0400, Gabriel Somlo wrote: > > When using `PLATFORM=generic` defaults, the kernel is loaded at > > `FW_JUMP_ADDR = FW_TEXT_START + 0x200000`, and the FDT is loaded at > > `FW_JUMP_FDT_ADDR = FW_TEXT_START + 0x2200000`. > > > > Therefore, the maximum kernel size before `FW_JUMP_FDT_ADDR` must > > be increased is the difference, or `0x2000000`. > > Sounds good, but this is only correct for RV64. For RV32 it should be > 0x1e00000. I wonder if we should do something like > > ``` > ${CROSS_COMPILE}objdump -h $KERNEL_ELF | sort -k 5,5 | awk -n '/^ +[0-9]+ /\ > {addr="0x"$3; size="0x"$5; printf "0x""%x\n",addr+size}' \ > - | (( `tail -1` > 0x2200000 )) && echo fdt overlaps kernel,\ > - increase FW_JUMP_FDT_ADDR > + | (( `tail -1` > (FW_JUMP_FDT_ADDR - FW_JUMP_ADDR) )) && echo fdt overlaps \ > + kernel, increase FW_JUMP_FDT_ADDR > > ${LLVM}objdump -h --show-lma $KERNEL_ELF | sort -k 5,5 | \ > awk -n '/^ +[0-9]+ / {addr="0x"$3; size="0x"$5; printf "0x""%x\n",addr+size}'\ > - | (( `tail -1` > 0x2200000 )) && echo fdt overlaps kernel,\ > - increase FW_JUMP_FDT_ADDR > + | (( `tail -1` > (FW_JUMP_FDT_ADDR - FW_JUMP_ADDR) )) && echo fdt overlaps \ > + kernel, increase FW_JUMP_FDT_ADDR > ``` > > instead? Yeah, v2 (with a commit blurb updated to match) should be on its way. Thanks, --Gabriel
diff --git a/docs/firmware/fw_jump.md b/docs/firmware/fw_jump.md index 956897e..d1b07b9 100644 --- a/docs/firmware/fw_jump.md +++ b/docs/firmware/fw_jump.md @@ -48,12 +48,12 @@ follows: ``` ${CROSS_COMPILE}objdump -h $KERNEL_ELF | sort -k 5,5 | awk -n '/^ +[0-9]+ /\ {addr="0x"$3; size="0x"$5; printf "0x""%x\n",addr+size}' \ - | (( `tail -1` > 0x2200000 )) && echo fdt overlaps kernel,\ + | (( `tail -1` > 0x2000000 )) && echo fdt overlaps kernel,\ increase FW_JUMP_FDT_ADDR ${LLVM}objdump -h --show-lma $KERNEL_ELF | sort -k 5,5 | \ awk -n '/^ +[0-9]+ / {addr="0x"$3; size="0x"$5; printf "0x""%x\n",addr+size}'\ - | (( `tail -1` > 0x2200000 )) && echo fdt overlaps kernel,\ + | (( `tail -1` > 0x2000000 )) && echo fdt overlaps kernel,\ increase FW_JUMP_FDT_ADDR ```
When using `PLATFORM=generic` defaults, the kernel is loaded at `FW_JUMP_ADDR = FW_TEXT_START + 0x200000`, and the FDT is loaded at `FW_JUMP_FDT_ADDR = FW_TEXT_START + 0x2200000`. Therefore, the maximum kernel size before `FW_JUMP_FDT_ADDR` must be increased is the difference, or `0x2000000`. Update the example calculation to reflect this, and avoid confusion. Signed-off-by: Gabriel Somlo <gsomlo@gmail.com> --- docs/firmware/fw_jump.md | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)