IMEM initialization error

[stdefeber]

Hello,

I have a relatively simple NEORV32 with the following options:

mem_int_imem_en => true
mem_int_imem_size => 64 * 1024
mem_int_dmem_en => true
mem_int_dmem_size => 64 * 1024
riscv_isa_c => true
riscv_isa_m => true
int_bootloader_en => false

However I get this error:

declaration of a too large object (256 kB > --max-stack-alloc=128 kB)

While the application image is below this 64KB size and below mentioned limits above.

object	size (bytes)
main.elf	61108
main.bin	20320
neorv32_application_image.vhd	66551

I can not find where this error is thrown.

cpusub(rtl).u_cpusub@cpusub(rtl).u_neorv32@neorv32_top(neorv32_top_rtl).neorv32_bus_gateway_inst@neorv32_bus_gateway(neorv32_bus_gateway_rtl).bus_monitor
  from: neorv32.neorv32_package.mem32_init_f
  from: neorv32.neorv32_imem(neorv32_imem_rtl).DECL_ELAB
  from: neorv32.neorv32_imem.LASTARCH.DECL_ELAB
  from: neorv32.neorv32_top(neorv32_top_rtl).memory_system.B1.neorv32_int_imem_inst_true.B1.neorv32_int_imem_inst.CMP_ELAB
  from: neorv32.neorv32_top(neorv32_top_rtl).DECL_ELAB
  from: neorv32.neorv32_top.LASTARCH.DECL_ELAB
  from: work.cpusub(rtl).u_neorv32.CMP_ELAB
  from: work.cpusub(rtl).DECL_ELAB
  from: work.cpusub.LASTARCH.DECL_ELAB
  from: work.tb_cpusub(rtl).u_cpusub.CMP_ELAB
  from: work.tb_cpusub(rtl).DECL_ELAB
  from: Elaboration of design

The application code is compiled separately.
While the VHDL is handled via ghdl.

br,

Simon

[stnolting]

Hey @stdefeber!

declaration of a too large object (256 kB > --max-stack-alloc=128 kB)

This happens when you invoke GHDL? Which version of GHDL are you using?

I've tested your configuration on my side and I cannot reproduce the error. However, you could try to increase the default GHDL / process stack size: https://ghdl.github.io/ghdl-cosim/vhpidirect/notebook/mistakes.html

• ulimit -s unlimited (to maximize the stack size of the process)
• --max-stack-alloc=0 (to maximize GHDL's internal stack)

[stdefeber]

Thanks for the quick reply.

I am using:

GHDL 4.1.0 (tarball) [Dunoon edition]
Compiled with GNAT Version: 13.3.0
GCC 13.3.0 code generator

I will try your suggestion for ulimit and max-stack allocation.

[stdefeber]

Issue is solved.

I had to figure out how to add the additional run options since I am using FuseSoc as build tool.

[stnolting]

Great to hear!