loom: corrupt when running urbit 3.0-65f03b3, upgrading from 2.12

[yosoyubik]

Print of the crash here:

urbit@yosoyubik:~$ ./norsyr-torryn/.run next --loom 32
~
urbit 2.12
boot: home is /home/urbit/norsyr-torryn
loom: mapped 2048MB
lite: arvo formula 2a2274c9
lite: core 4bb376f0
lite: final state 4bb376f0
loom: mapped 4096MB
boot: protected loom
live: mapped: GB/3.452.125.184
live: loaded: KB/16.384
boot: installed 661 jets
vere: next (%edge): 3.0-65f03b3
vere: saved to /home/urbit/norsyr-torryn/.bin/edge/vere-v3.0-65f03b3-linux-x86_64
vere: upgrade succeeded
urbit@yosoyubik:~$ ./norsyr-torryn/.run -p 32124 --loom 32
~
urbit 3.0-65f03b3
boot: home is /home/urbit/norsyr-torryn
disk: loading old format
loom: mapped 4096MB
boot: protected loom
live: mapped: GB/3.452.125.184
live: loaded: KB/16.384
loom: memoization migration running...
Assertion '!"loom: corrupt"' failed in pkg/noun/allocate.c:185

bail: oops
home: bailing out
Aborted (core dumped)

@matthew-levan

[yosoyubik]

Running with --no-demand fixed it. One of the issues (since I got a different crash than the one reported here) seemed to be that I had exceed the system-wide limit on mappings:

~/vere$ cat /proc/sys/vm/max_map_count
65530
~/vere$ cat /proc/5238/maps | wc -l
65531

(Increasing the number of mappings also fixes it: sudo sysctl -w vm.max_map_count=262144)

[joemfb]

For posterity, the issue fixed by --no-demand or increasing the vm mapping limit is mprotect: fail: Out of memory. I still don't know why you were initially seeing loom: corrupt assertions, or why they went away.