This seems to happen starting from the initial integration of JDK-8264207. I believe I only see it on my desktop, because there hsdis is in /usr/lib, and so the hotspot disassembler is enabled for all VMs, including the test one. If I remove that system-wide hsdis, the test starts to pass.

Actually, I think the test path in disasm_string_check is not taken for a default system installation, as no hsdis is present in test JDK.

$ CONF=linux-x86_64-server-fastdebug make run-test TEST=gtest:codestrings
Building target 'run-test' in configuration 'linux-x86_64-server-fastdebug'
Updating hotspot/variant-server/libjvm/gtest/libjvm.so due to 1 file(s)
Test selection 'gtest:codestrings', will run:
* gtest:codestrings/server

Running test 'gtest:codestrings/server'
Note: Google Test filter = codestrings*
[==========] Running 1 test from 1 test case.
[----------] Global test environment set-up.
[----------] 1 test from codestrings
[ RUN ] codestrings.validate_vm
/home/shade/trunks/jdk/test/hotspot/gtest/code/test_codestrings.cpp:81: Failure
Expected equality of these values:
  replace_addr_expr(out1.as_string())
    Which is: "--------------------------------------------------------------------------------\n ;; First block comment.\n <addr>: nop\n--------------------------------------------------------------------------------\n"
  replace_addr_expr(out2.as_string())
    Which is: "--------------------------------------------------------------------------------\n ;; First block comment.\n <addr>: nop\n <addr>: hlt \n <addr>: hlt \n <addr>: hlt \n <addr>: hlt \n <addr>: hlt \n <addr>: hlt \n <addr>: hlt \n--------------------------------------------------------------------------------\n"
With diff:
@@ +2,10 @@
  ;; First block comment.
   <addr>: nop
+ <addr>: hlt
+ <addr>: hlt
+ <addr>: hlt
+ <addr>: hlt
+ <addr>: hlt
+ <addr>: hlt
+ <addr>: hlt
 --------------------------------------------------------------------------------\n