run time bug in the gcc command line test

[alanjhewitt]

I am getting a run time SIGSEGV error at a specific allocate statement in my branch. This problem only occurs for the gcc build - not the ifort and not for the rose-stem either.

At revision 1439, this occured at the allocate statement on line 3444.

I didnt feel like there was anything wrong with the allocate, so i moved a whole block of code up the kernel, and it ran past that problem allocate to the next allocate (now on line 3438).
https://code.metoffice.gov.uk/trac/lfric_apps/changeset/1446/main/branches/dev/alanjhewitt/um13.5_ukca_3d

I thought there might be some available memory issue, so I tried with ulimit -s unlimited, but had the same error.

The error output is as follows:

0240520130155.079+0100:S1:INFO : environ_fullhtp1_real - after step 1)
20240520130155.079+0100:S1:INFO : seg_len = 4
20240520130155.079+0100:S1:INFO : nlayers = 70
20240520130155.079+0100:S1:INFO : m_fields = 1
20240520130155.079+0100:S1:INFO : environ_fullhtp1_real - after step 2)
20240520130155.079+0100:S1:INFO : after allocate(environ_fullhtp1_real)
20240520130155.079+0100:S1:INFO : before environ_fullhtp1_real
20240520130155.079+0100:S1:INFO : after environ_fullhtp1_real

Program received signal SIGSEGV: Segmentation fault - invalid memory reference.

Backtrace for this error:
#0 0x2b4c06d3762f in ???
#1 0x2b4c06fc67b4 in ???
#2 0x2b4c06fc978b in ???
#3 0x11ea657 in __aerosol_ukca_kernel_mod_MOD_aerosol_ukca_code
at kernel/aerosol_ukca_kernel_mod.F90:3438
#4 0xddb2ff in __aerosol_ukca_alg_mod_psy_MOD_invoke_0_aerosol_ukca_kernel_type
at algorithm/aerosol_ukca_alg_mod_psy.f90:912
#5 0xdb089f in __aerosol_ukca_alg_mod_MOD_aerosol_ukca_alg
at algorithm/aerosol_ukca_alg_mod.f90:548
#6 0x859975 in __semi_implicit_timestep_alg_mod_MOD_run_step
at algorithm/semi_implicit_timestep_alg_mod.f90:497
#7 0x85f8df in __semi_implicit_timestep_alg_mod_MOD_semi_implicit_alg_step
at algorithm/semi_implicit_timestep_alg_mod.f90:199
#8 0x543d21 in __gungho_step_mod_MOD_gungho_step
at driver/gungho_step_mod.f90:88
#9 0x52694a in __gungho_driver_mod_MOD_step
at driver/gungho_driver_mod.F90:302
#10 0x520f6b in lfric_atm
at /net/data/users/ahewitt/branches/LFRic4333/um13.5_ukca_3d/applications/lfric_atm/working/build_lfric_atm/lfric_atm.f90:70
#11 0x521083 in main
at /net/data/users/ahewitt/branches/LFRic4333/um13.5_ukca_3d/applications/lfric_atm/working/build_lfric_atm/lfric_atm.f90:19
Segmentation fault
[ahewitt@vld673:/data/users/ahewitt/branches/LFRic4333/um13.5_ukca_3d/applications/lfric_atm/example]$

https://code.metoffice.gov.uk/trac/lfric_apps/log/main/branches/dev/alanjhewitt/um13.5_ukca_3d
https://code.metoffice.gov.uk/trac/ukca/log/main/branches/dev/alanjhewitt/um13.5_lfric_ukca_3d

[alanjhewitt]

the rose-stem and ifort command line tests do not run to completion yet. both fail with an issue around the frac_land field. i am pretty sure i have made an indexing error in my branch. i think the two issues are separate, and am trying to solve them in parallel, but it is possible that this SIGSEGV will go away if i fix the other problem.