mMailboxPointer is used before it is initialization. This issue only happens
when SmmDebugAgent is initialized without PeiDebugAgent or DxeDebugAgent
initialization. The fix is to use Mailbox instead.
Cc: Ruiyu Ni <ruiyu.ni@intel.com>
Cc: Hao Wu <hao.a.wu@intel.com>
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Jeff Fan <jeff.fan@intel.com>
Reviewed-by: Ruiyu Ni <ruiyu.ni@intel.com>
DxeDebugAgentLib instance produces gEfiDebugAgentGuid system table.
Its usage should be PRODUCES instead of SOMETIMES_PRODUCES.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Liming Gao <liming.gao@intel.com>
Reviewed-by: Jeff Fan <jeff.fan@intel.com>
git-svn-id: https://svn.code.sf.net/p/edk2/code/trunk/edk2@19316 6f19259b-4bc3-4df7-8a09-765794883524
Actually, TimerCycle is APIC timer's initial count. Timer count value 0 is
missed when calculating Delta value.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Jeff Fan <jeff.fan@intel.com>
Reviewed-by: Ruiyu Ni <ruiyu.ni@intel.com>
git-svn-id: https://svn.code.sf.net/p/edk2/code/trunk/edk2@18329 6f19259b-4bc3-4df7-8a09-765794883524
If CPU runs fast and timer runs slow, two GetApicTimerCurrentCount() may return
the same timer count value. We need to consider timer roll-over not happened.
Otherwise, one false timeout flag will be set.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Jeff Fan <jeff.fan@intel.com>
Reviewed-by: Ruiyu Ni <ruiyu.ni@intel.com>
git-svn-id: https://svn.code.sf.net/p/edk2/code/trunk/edk2@18328 6f19259b-4bc3-4df7-8a09-765794883524
In DEBUG_AGENT_INIT_POSTMEM_SEC case, caller may disable/restore CPU interrupt
to protect the stack/heap migration. SecPeiDebugAgentLib cannot always enable
CPU interrupt. Otherwise system may crash during stack/heap migration.
SecPeiDebugAgentLib should restore original CPU interrupt state in
DEBUG_AGENT_INIT_POSTMEM_SEC case.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Jeff Fan <jeff.fan@intel.com>
Reviewed-by: Brian J. Johnson <bjohnson@sgi.com>
Tested-by: Brian J. Johnson <bjohnson@sgi.com>
git-svn-id: https://svn.code.sf.net/p/edk2/code/trunk/edk2@17916 6f19259b-4bc3-4df7-8a09-765794883524
Recent changes to debug timer initialization (commit 2befbc82, svn 17572)
modified the Sec/Pei InitializeDebugAgent() routine to enable debug timer
interrupts. This causes problems in the DEBUG_AGENT_INIT_POSTMEM_SEC case:
the callers appear to assume that if they block timer interrupts before the
call, interrupts will remain blocked afterwards.
It is not always safe to have interrupts enabled on return from
InitializeDebugAgent(). For instance, after calling InitializeDebugAgent(),
OvmfPkg's TemporaryRamMigration() moves the stack, heap, and IDT to RAM, then
switches to the new stack. Only then does it reenable timer interrupts.
Taking an interrupt during this process can corrupt state, causing crashes.
Do not unmask the debug timer interrupt in the DEBUG_AGENT_INIT_POSTMEM_SEC
case.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Brian J. Johnson <bjohnson@sgi.com>
Reviewed-by: Jeff Fan <jeff.fan@intel.com>
git-svn-id: https://svn.code.sf.net/p/edk2/code/trunk/edk2@17893 6f19259b-4bc3-4df7-8a09-765794883524
Recent changes to debug timer handling ended up leaving the timer disabled in
PEI. This made it impossible to stop execution in PEI externally via the
debugger. Enable the timer when InitializeDebugAgent calls
InitializeDebugAgentPhase2, as well as when it returns directly.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Brian J. Johnson <bjohnson@sgi.com>
Reviewed-by: Jeff Fan <jeff.fan@intel.com>
git-svn-id: https://svn.code.sf.net/p/edk2/code/trunk/edk2@17819 6f19259b-4bc3-4df7-8a09-765794883524
OffsetHigh is 16bit value and its type is UINT32 and defined in structure.
It will be 32bit int type after do 16-bit left-shift operation. Then it will
sign extension if cast it to UINT64 if its high bit is 1.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Jeff Fan <jeff.fan@intel.com>
Reviewed-by: Ruiyu Ni <ruiyu.ni@intel.com>
Reviewed-by: Scott Duplichan <scott@notabs.org>
git-svn-id: https://svn.code.sf.net/p/edk2/code/trunk/edk2@17716 6f19259b-4bc3-4df7-8a09-765794883524
Now Debug Agent library uses Local APIC Timer to implement time-out mechanism.
For AP, its local APIC timer may not work. This fix is to initialize Local
APIC timer if it doesn't work as expected when debugging AP function.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Jeff Fan <jeff.fan@intel.com>
Reviewed-by: Ruiyu Ni <ruiyu.ni@intel.com>
git-svn-id: https://svn.code.sf.net/p/edk2/code/trunk/edk2@17604 6f19259b-4bc3-4df7-8a09-765794883524
InitializeApicTimer() will enable Local APIC timer interrupt. Even though we
disable CPU interrupt at the beginning and enable CPU Interrupt after debug
agent initialized completely, some Boot Service may invoke RestoreTpl () which
may enable CPU interrupt.
We should disable Local APIC timer in InitializeDebugTimer () to avoid Debug
Timer interrupt happens during debug port and debug agent initialization phase.
And enable Debug Timer interrupt after debug agent is initialized.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Jeff Fan <jeff.fan@intel.com>
Reviewed-by: Ruiyu Ni <ruiyu.ni@intel.com>
git-svn-id: https://svn.code.sf.net/p/edk2/code/trunk/edk2@17572 6f19259b-4bc3-4df7-8a09-765794883524
Check PcdDebugPortHandleBufferSize before allocate buffer.
PeriodicMode is BOOLEAN type, needn't to use == TRUE in if condition.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Jeff Fan <jeff.fan@intel.com>
Reviewed-by: Jaben Carsey <jaben.carsey@intel.com>
git-svn-id: https://svn.code.sf.net/p/edk2/code/trunk/edk2@17558 6f19259b-4bc3-4df7-8a09-765794883524
Now Debug Agent library uses Local APIC Timer to implement time-out mechanism.
In SMM, SMM BSP maybe not be the one in DXE phase, its local APIC timer may not
work. This fix is to initialize Local APIC timer if it doesn't work as expected
at SMM entry.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Jeff Fan <jeff.fan@intel.com>
Reviewed-by: Ruiyu Ni <ruiyu.ni@intel.com>
git-svn-id: https://svn.code.sf.net/p/edk2/code/trunk/edk2@17500 6f19259b-4bc3-4df7-8a09-765794883524
The root cause is SerialPortDebugCommunicationLib returns NULL debug port handle causing the CompressData() doesn't send the compressed data.
The fix doesn't rely on the debug port handle value and uses Send parameter to indicate whether to send the compressed data.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Ruiyu Ni <ruiyu.ni@intel.com>
Reviewed-by: Jeff Fan <jeff.fan@intel.com>
git-svn-id: https://svn.code.sf.net/p/edk2/code/trunk/edk2@17464 6f19259b-4bc3-4df7-8a09-765794883524
Now DxeDebugAgent Library instance will print help information on how to load
DebugAgentDxe.efi in UEFI shell. But it is printed after Target connected to
Host side. This fix is to move help info print to DebugAgentDxe module before
Target tries to connect HOST. It could help developer to get useful information
as early as possible.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Jeff Fan <jeff.fan@intel.com>
Reviewed-by: Ruiyu Ni <ruiyu.ni@intel.com>
git-svn-id: https://svn.code.sf.net/p/edk2/code/trunk/edk2@17458 6f19259b-4bc3-4df7-8a09-765794883524
Jeff Fan
Laszlo Ersek
Eric Dong
Hao Wu
Liming Gao
Jordan Justen
Michael Kinney
Ruiyu Ni
Qiu Shumin
Brian J. Johnson