crash-utility
commented
5 years ago When you see that error, there should have been a dump of the memory allocation statistics just before the message. You can also see them at any time during runtime by entering "help -b":
crash> help -b buf_1K_used: 815 buf_2K_used: 67 buf_4K_used: 29 buf_8K_used: 10 buf_32K_used: 100 buf_1K_ovf: 0 buf_2K_ovf: 0 buf_4K_ovf: 0 buf_8K_ovf: 0 buf_32K_ovf: 0 buf_1K_maxuse: 2 of 10 buf_2K_maxuse: 1 of 10 buf_4K_maxuse: 1 of 5 buf_8K_maxuse: 1 of 5 buf_32K_maxuse: 1 of 1 buf_inuse[5]: [0][0][0][0][0] smallest: 16 largest: 1740437 embedded: 0 max_embedded: 3 mallocs: 50 frees: 50 reqs/total: 1071/12356581 average size: 11537 crash>
There are 31 static buffers that get used first, and if those are exhausted, it starts using calloc(). It will fail, dump the stats above, and print the error message you are seeing if:
(1) it calls calloc() more than 2000 times, or (2) if a calloc() call fails.
The buffers are for temporary usage. After session intialization and prior to the first "crash> " prompt, and then after each runtime command, all of the buffers will have been freed. So in the example above, which was taken after initialization, note that the bp_inuse[ ] counters are all 0, and the mallocs/frees counts are equal. That would be true during the whole crash session.
So if (1) were true, then the "mallocs:" and "frees:" counts above would be totally out of whack, where normally they would be equal. That's highly unlikely, so presumably (2) occurred, but it's unclear when and where one of the calloc() calls failed. If you run crash with a debug count on the command line, say with "-d4", it may illuminate where the intialization code was running when it gave up.
jpfreyen
This is in crash 7.2.1-2 and from compiling the source from 'master' branch (pulled today)
I get the 'crash: cannot allocate any more memory!' error. I have 8GB and 2 CPUs dedicated to my Fedora VM, I thought that would be plenty of memory as when I start up crash I don't see 8GB being used up in system monitor.
I searched around with Google and it seems like when this error shows up its associated with a bugzilla entry, so maybe this is a real problem with crash? I certainly cannot find any suggested command-line startups to use to try and work around this.
All I am doing is typing 'crash' on command-line and hitting enter. I was trying to use:
'crash vmlinux vmcore'
in which the vmcore is 214MB but I get that same error regardless if I type 'crash' on the command-line or try and load a vmcore file.