Manifests with recently added test on x86_32:

$ CONF=linux-x86-server-fastdebug make test TEST=serviceability/sa/ClhsdbThreadContext.java

...

Thread "SteadyStateThread" id=2303604 Address=0xa52d32e0
GS: 0x00000063
FS: null
ES: 0x0000002b
DS: 0x0000002b
EDI: null
ESI: 0xa3e7cb40: In java stack [0xa3e7e000,null,0xa3e2e000] for thread sun.jvm.hotspot.runtime.JavaThread@0xa52d32e0:
   "SteadyStateThread" #19 prio=5 tid=0xa52d32e0 nid=2303604 waiting for monitor entry [0xa3e7c000]
   java.lang.Thread.State: BLOCKED (on object monitor)
   JavaThread state: _thread_blocked
EBP: 0xffffffff
ESP: null
EBX: 0xa52d3a68
EDX: null
ECX: 0x00000089
EAX: 0xfffffdfc
TRAPNO: null
ERR: null
EIP: 0xf7f38549
CS: 0x00000023
EFLAGS: null
UESP: 0xa3e7cb08: In java stack [0xa3e7e000,null,0xa3e2e000] for thread sun.jvm.hotspot.runtime.JavaThread@0xa52d32e0:
   "SteadyStateThread" #19 prio=5 tid=0xa52d32e0 nid=2303604 waiting for monitor entry [0xa3e7c000]
   java.lang.Thread.State: BLOCKED (on object monitor)
   JavaThread state: _thread_blocked


java.lang.RuntimeException: Test ERROR java.lang.RuntimeException: 'In java stack \[0x\p{XDigit}+,0x\p{XDigit}+,0x\p{XDigit}+\] for thread' missing from stdout/stderr
```

The test fails because debugger output prints `null` in `[0xa3e7e000,null,0xa3e2e000]`, which is read from ESP. However, we actually populate only `SP`, which is the alias to `UESP`, *not* `ESP`, in `Java_sun_jvm_hotspot_debugger_linux_LinuxDebuggerLocal_getThreadIntegerRegisterSet0`. See the snippet above: `UESP` is definitely there.

`UESP` is user stack pointer in Solaris, `sp_at_signal` in Linux. It looks to me from Linux sources that `sp_at_signal` is always populated with the same value as `sp`. So we either need to populate `ESP` from `get_lwp_regs` input, or we should pull from `SP` (which is aliased to `UESP`) in `*X86JavaThreadPDAccess.getLastSP()`. Prior work in JDK-8208091 did the latter.