[LU-17405] Executable created with gcc gives ELF interpreter error (2.15 client w/ 2.12 server) Created: 08/Jan/24  Updated: 31/Jan/24

Status: Open
Project: Lustre
Component/s: None
Affects Version/s: Lustre 2.15.2
Fix Version/s: None

Type: Bug Priority: Minor
Reporter: Mike D Assignee: WC Triage
Resolution: Unresolved Votes: 0
Labels: None
Environment:

Lustre 2.15.2.2 client with 2.12.6 server. CentOS 7 (various kernels)


Issue Links:
Related
is related to LU-17323 fork() leaks ERESTARTNOINTR (errno 51... Open
Severity: 3
Rank (Obsolete): 9223372036854775807

 Description   

I noticed this problem while testing Lustre 2.15 client on a system with 2.12.6 servers.

The behavior is different with csh and bash.  With csh, ./a.out isn't there until I run the ls.  With bash, it runs after the ELF interpreter error.  This is reproducible every time.

 

  • csh case (a.out not found until the ls is run)

$ gcc hello.c

$ ./a.out

./a.out: Command not found

$ ./a.out

./a.out: Command not found

$ ./a.out

./a.out: Command not found

$ ls a.out

a.out

$ ./a.out

hello world

 

  • bash case (a.out works after the bad ELF interpreter error)

$ gcc hello.c

$ ./a.out

bash: ./a.out: /lib64/ld-linux-x86-64.so.2: bad ELF interpreter: No such file or directory

$ ./a.out

hello world

 

The main thing I've noticed in the client debug log:

running a.out

file.c:2012:ll_file_read_iter() file a.out:[0x2000XXXXX:0x9:0x0], ppos: 0, count: 80

file.c:2012:ll_file_read_iter() file a.out:[0x2000XXXXX:0x9:0x0], ppos: 6456, count: 1984

file.c:2012:ll_file_read_iter() file a.out:[0x2000XXXXX:0x9:0x0], ppos: 6186, count: 268

file.c:2012:ll_file_read_iter() file a.out:[0x2000XXXXX:0x9:0x0], ppos: 64, count: 504

file.c:2012:ll_file_read_iter() file a.out:[0x2000XXXXX:0x9:0x0], ppos: 568, count: 200

running a.out again

file.c:2012:ll_file_read_iter() file a.out:[0x2000XXXXX:0x9:0x0], ppos: 0, count: 128

file.c:2012:ll_file_read_iter() file a.out:[0x2000XXXXX:0x9:0x0], ppos: 64, count: 504

file.c:2012:ll_file_read_iter() file a.out:[0x2000XXXXX:0x9:0x0], ppos: 568, count: 28



 Comments   
Comment by Mike D [ 31/Jan/24 ]

I would probably classify this as a major bug.  I probably should have set it to major.  This issue is easily reproducible and is a blocker for any future upgrades.  

Generated at Sat Feb 10 03:35:10 UTC 2024 using Jira 9.4.14#940014-sha1:734e6822bbf0d45eff9af51f82432957f73aa32c.