07:18 <ysionneau> mwalle: I need to confirme this but, I think I have an issue with the kernel loading in qemu

07:19 <ysionneau> don't know if it's my ELF's fault or qemu's fault

07:54 <ysionneau> I have a symbol (in .data I think) named v_putc which is a function pointer, if I objdump -D the ELF binary I can see it is initialized with a value (which is the putc() of the milkymist uart driver)

07:55 <ysionneau> but if I x/1x v_putc I get 0x00000000

07:55 <ysionneau> in gdb

07:55 <ysionneau> at startup

07:55 <ysionneau> well, on the symbol name it's normal I get all 0 because I would be reading at the virtual address then

07:56 <ysionneau> but if I translate and read the 0x40****** address I get all 0 as well

07:57 <ysionneau> anyway, I won't have the time to work on this until sunday

11:13 <mwalle> ysionneau: (1) please send me the binary. (2) was it working with your hacked qemu and the objcopy -Obinary method?

11:16 <ysionneau> (1) I will try to send it this afternoon but not sure, maybe on sunday :/

11:16 <ysionneau> (2) yes it was working, it was sending characters to the uart

11:17 <ysionneau> I did not specifically print the v_putc value however

11:17 <ysionneau> since it was working I did not try to "debug" it

16:35 <ysionneau> mwalle: qemu is fine =) (you would have bet I guess)

16:35 <ysionneau> it's my ELF virtual memory which is wrong

16:47 <ysionneau> hum my linker script is wrong in the email

16:50 <ysionneau> but fixing it does not change the ELF layout

17:17 <ysionneau> I imagine that the linker does that to align different sections with different permissions on a 4kB page boundary

17:17 <ysionneau> but it's a pain for the early *tlb miss handlers

17:18 <ysionneau> which are just doing TLBPADDR = virt_to_phy(TLBPADDR);

17:18 <ysionneau> virt_to_phy() is now a pain to implement

17:19 <ysionneau> for itlb miss it's OK though

17:25 <ysionneau> *bbl*

18:00 <mumptai> moin

20:14 <sb0> hi

20:39 <mwalle> ho

21:10 <ysionneau> hi