Cannot run an enclave from an eif that requires ~150gb

[Nesopie]

Initialy, I couldn't allocate ~150 gb of unfragmented memory to the nitro allocator service on a single numa node so I had to edit my grub config to allocate 200gb on a single numa node (I'm using a c7a.48large instance). However, after running my enclave these are the logs I get:

`Could not open /env file: No such file or directory

Could not open /env file: No such file or directory
[ 36.574060] init[1]: segfault at 0 ip 00000000004089a4 sp 00007ffeb8acc9d0 error 4 in init[400000+b2000]
[ 36.575169] Kernel panic - not syncing: Attempted to kill init! exitcode=0x0000000b
[ 36.575169]
[ 36.576200] CPU: 2 PID: 1 Comm: init Tainted: G OE 4.14.256-209.484.amzn2.x86_64 #1
[ 36.577155] Call Trace:
[ 36.577432] dump_stack+0x6d/0x8b
[ 36.577800] panic+0xdf/0x242
[ 36.578136] ? do_coredump+0x743/0xf70
[ 36.578542] do_exit+0x7d6/0xb80
[ 36.578892] do_group_exit+0x3e/0xb0
[ 36.579279] get_signal+0x11b/0x6d0
[ 36.579667] do_signal+0x32/0x610
[ 36.580042] ? __bad_area_nosemaphore+0xd2/0x1b0
[ 36.580529] ? printk+0x4d/0x69
[ 36.580859] ? __do_page_fault+0x345/0x4b0
[ 36.581276] exit_to_usermode_loop+0x52/0x90
[ 36.581719] ? async_page_fault+0x2f/0x50
[ 36.582133] prepare_exit_to_usermode+0x64/0x70
[ 36.582606] retint_user+0x8/0x8
[ 36.582946] RIP: 0033:0x4089a4
[ 36.583270] RSP: 002b:00007ffeb8acc9d0 EFLAGS: 00010202
[ 36.583821] RAX: 0000000000000000 RBX: 0000000000000000 RCX: 0000000000000000
[ 36.584554] RDX: 00000000006b4650 RSI: 0000000000000034 RDI: 0000000000000000
[ 36.585285] RBP: 0000000000000000 R08: 0000000000000019 R09: 000000000049c6c0
[ 36.586040] R10: 00000000ffffffe7 R11: 0000000000000246 R12: 0000000000402020
[ 36.586803] R13: 0000000001015840 R14: 0000000000000000 R15: 0000000000000000
[ 36.587810] Kernel Offset: disabled
[ 36.588255] Rebooting in 30 seconds..
[ E45 ] Enclave console read error. Such error appears when reading from a running enclave's console fails.
For more details, please visit https://docs.aws.amazon.com/enclaves/latest/user/cli-errors.html#E45
If you open a support ticket, please provide the error log found at "/var/log/nitro_enclaves/err2025-02-25T06:55:44.952765148+00:00.log"`

[atanzu]

Hi @Nesopie
Could you please share more info on the topic:

• When you run an enclave, did you allocate all those 200 GiB, and does the issue reproduce if you allocate less? Also how many vCPUs do you use?
• Could you please share your GRUB config and Nitro Allocator config?
• Could you please share the error log from /var/log?

[Strnadj]

Hey @atanzu,

we are having the same issue...

• Arguments
  --enclave-cid 16 --cpu-count 4 --memory 60000 --eif-path (minimum by cli is 19000)

• Current memory:

free -h
              total        used        free      shared  buff/cache   available
Mem:           123G         63G         55G        1.3M        5.3G         59G

• config:

---
# Enclave configuration file.
#
# How much memory to allocate for enclaves (in MiB).
memory_mib: 64000
#
# How many CPUs to reserve for enclaves
cpu_count: 16

• There is no information in the log, but a few lines before the could not open /ENV is this:

[    0.844929] NET: Registered PF_INET6 protocol family
[    3.355166] Initramfs unpacking failed: write error
[    9.396625] Freeing initrd memory: 11872984K

cat /var/log/nitro_enclaves/err2025-09-06T05:45:38.621058913+00:00.log
  Action: Enclave Console
  Subactions:
    Failed to connect to enclave console
    Connect to enclave console
    Failed to read data from the console: ENOTCONN
  Root error file: src/utils.rs
  Root error line: 229
  Version: 1.4.2

[Strnadj]

Another discovery from me, that kernels distributed together with cli-devel should be this one:

ed24913346a34d719afa2031299253160a2e3460

This one works... I can do 100GB with no issues; it took significantly more time, but the one from the main did not work.

But