Skip to contentSkip to conversation

kdb: Avoid using dbg_io_ops until it is initialized

PATCH In review Tim Bird 3 participants 5 messages Sep 21, 2011 – Mar 20, 2012 KGDB / KDB /debug_core

lore
Share & tools
lore

Tim Bird

author
#0 lore raw

PATCH 1/1 kdb: Avoid using dbg_io_ops until it is initialized

This fixes a bug with setting a breakpoint during kdb initialization (from kdb_cmds). Any call to kdb_printf() before the initialization of the kgdboc serial console driver (which happens much later during bootup than kdb_init), results in kernel panic due to the use of dbg_io_ops before it is initialized.

Signed-off-by Tim Bird <[email protected]>
---
 kernel/debug/kdb/kdb_io.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)
kernel/debug/kdb/kdb_io.c +1 -1
index c9b7f4f..3bc995f 100644
@@ -675,7 +675,7 @@ kdb_printit:
 	if (!dbg_kdb_mode && kgdb_connected) {
 		gdbstub_msg_write(kdb_buffer, retlen);
 	} else {
-		if (!dbg_io_ops->is_console) {
+		if (dbg_io_ops && !dbg_io_ops->is_console) {
 			len = strlen(kdb_buffer);
 			cp = kdb_buffer;
 			while (len--) {
Signature · 1 line
1.7.2.3
#1 lore raw
Quoted 5 lines · Tim Bird

This fixes a bug with setting a breakpoint during kdb initialization (from kdb_cmds).  Any call to kdb_printf() before the initialization of the kgdboc serial console driver (which happens much later during bootup than kdb_init), results in kernel panic due to the use of dbg_io_ops before it is initialized.

Did you get the panic for that case?

Dongdong

Quoted 35 lines
Signed-off-by Tim Bird <[email protected]>

---
 kernel/debug/kdb/kdb_io.c |    2 +-

 1 files changed, 1 insertions(+), 1 deletions(-)
kernel/debug/kdb/kdb_io.c +1 -2
index c9b7f4f..3bc995f 100644
@@ -675,7 +675,7 @@ kdb_printit:
        if (!dbg_kdb_mode && kgdb_connected) {
                gdbstub_msg_write(kdb_buffer, retlen);
        } else {
-               if (!dbg_io_ops->is_console) {
+               if (dbg_io_ops && !dbg_io_ops->is_console) {
                        len = strlen(kdb_buffer);
                        cp = kdb_buffer;
                        while (len--) {
--

1.7.2.3

------------------------------------------------------------------------------
All the data continuously generated in your IT infrastructure contains a definitive record of customers, application performance, security threats, fraudulent activity and more. Splunk takes this data and makes sense of it. Business sense. IT sense. Common sense.
http://p.sf.net/sfu/splunk-d2dcopy1
_______________________________________________
Kgdb-bugreport mailing list [email protected]
https://lists.sourceforge.net/lists/listinfo/kgdb-bugreport

author
#2 lore raw
Quoted 9 lines · Dongdong Deng
Quoted 5 lines · Tim Bird

This fixes a bug with setting a breakpoint during kdb initialization (from kdb_cmds). Any call to kdb_printf() before the initialization of the kgdboc serial console driver (which happens much later during bootup than kdb_init), results in kernel panic due to the use of dbg_io_ops before it is initialized.

Did you get the panic for that case?

Yes. See below (without the patch):

[    0.009552] BUG: unable to handle kernel NULL pointer dereference at 0000001c
[    0.009999] IP: [<c109112f>] vkdb_printf+0xef/0x690
[    0.009999] *pde = 00000000
[    0.009999] Oops: 0000 [#1] SMP
[    0.009999] Modules linked in:
[    0.009999]
[    0.009999] Pid: 0, comm: swapper Not tainted 3.1.0-rc2+ #10                  /D865PERL
[    0.009999] EIP: 0060:[<c109112f>] EFLAGS: 00010002 CPU: 0
[    0.009999] EIP is at vkdb_printf+0xef/0x690
[    0.009999] EAX: 00000000 EBX: c18d06c0 ECX: 0000000f EDX: 00000001
[    0.009999] ESI: 0000000f EDI: c184cd2a EBP: c18adee0 ESP: c18ade70
[    0.009999]  DS: 007b ES: 007b FS: 00d8 GS: 0000 SS: 0068
[    0.009999] Process swapper (pid: 0, ti=c18ac000 task=c18b1fa0 task.ti=c18ac000)
[    0.009999] Stack:
[    0.009999]  c18adef8 c18adeb8 c105ecc2 007a6bc8 00000000 00000202 209db643 00000000
[    0.009999]  c18adeb0 00000000 017f5762 c17f5764 c18adebc c18adf40 00000000 c19db643
[    0.009999]  c18adf08 c1094040 c17f5767 c18adf00 c18adecc c19db720 00000001 00000000
[    0.009999] Call Trace:
[    0.009999]  [<c105ecc2>] ? sched_clock_local+0xb2/0x190
[    0.009999]  [<c1094040>] ? kdbgetaddrarg+0x80/0x220
[    0.009999]  [<c10916de>] kdb_printf+0xe/0x10
[    0.009999]  [<c1097bc9>] kdb_printbp+0x39/0xc0
[    0.009999]  [<c111cd30>] ? sync_one_sb+0x20/0x20
[    0.009999]  [<c1097e25>] kdb_bp+0x1d5/0x260
[    0.009999]  [<c111cd30>] ? sync_one_sb+0x20/0x20
[    0.009999]  [<c1094ecd>] kdb_parse+0x2ed/0x630
[    0.009999]  [<c1097950>] ? kdb_bt+0x3b0/0x3b0
[    0.009999]  [<c1913208>] ? kdb_initbptab+0x188/0x18e
[    0.009999]  [<c191301b>] kdb_init+0x5d2/0x637
[    0.009999]  [<c19129f1>] dbg_late_init+0x22/0x24
[    0.009999]  [<c18fd7a3>] start_kernel+0x2a8/0x2ea
[    0.009999]  [<c18fd325>] ? kernel_init+0x13a/0x13a
[    0.009999]  [<c18fd0b3>] i386_start_kernel+0xb3/0xbb
[    0.009999] Code: 00 c6 45 ab 20 c7 45 ac 00 00 00 00 b8 e0 b3 9d c1 e8 76 7a 18 00 8b 15 44 b8 8b c1 85 d2 89 c6 0f 84 1
6 01 00 00 a1 74 65 9d c1 <8b> 78 1c 85 ff 75 19 e9 b3 02 00 00 90 8d 74 26 00 89 d8 89 f1
[    0.009999] EIP: [<c109112f>] vkdb_printf+0xef/0x690 SS:ESP 0068:c18ade70
[    0.009999] CR2: 000000000000001c
[    0.009999] ---[ end trace 4eaa2a86a8e2da22 ]---
[    0.009999] Kernel panic - not syncing: Attempted to kill the idle task!
[    0.009999] Pid: 0, comm: swapper Tainted: G      D     3.1.0-rc2+ #10
[    0.009999] Call Trace:
[    0.009999]  [<c160a2f7>] panic+0x57/0x157
[    0.009999]  [<c103ea2b>] do_exit+0x6ab/0x7a0
[    0.009999]  [<c103d2e7>] ? kmsg_dump+0x37/0xc0
[    0.009999]  [<c160a41f>] ? printk+0x28/0x2a
[    0.009999]  [<c160dc45>] oops_end+0x85/0xc0
[    0.009999]  [<c1023ede>] no_context+0xbe/0x150
[    0.009999]  [<c1024000>] __bad_area_nosemaphore+0x90/0x140
[    0.009999]  [<c1041887>] ? irq_exit+0x57/0x90
[    0.009999]  [<c10039d6>] ? do_IRQ+0x46/0xb0
[    0.009999]  [<c160f590>] ? spurious_fault+0xd0/0xd0
[    0.009999]  [<c10240c2>] bad_area_nosemaphore+0x12/0x20
[    0.009999]  [<c160f869>] do_page_fault+0x2d9/0x420
[    0.009999]  [<c121007b>] ? kobject_uevent_env+0x3bb/0x400
[    0.009999]  [<c12164d8>] ? string+0x38/0xc0
[    0.009999]  [<c160f590>] ? spurious_fault+0xd0/0xd0
[    0.009999]  [<c160d1d2>] error_code+0x5a/0x60
[    0.009999]  [<c160f590>] ? spurious_fault+0xd0/0xd0
[    0.009999]  [<c109112f>] ? vkdb_printf+0xef/0x690
[    0.009999]  [<c105ecc2>] ? sched_clock_local+0xb2/0x190
[    0.009999]  [<c1094040>] ? kdbgetaddrarg+0x80/0x220
[    0.009999]  [<c10916de>] kdb_printf+0xe/0x10
[    0.009999]  [<c1097bc9>] kdb_printbp+0x39/0xc0
[    0.009999]  [<c111cd30>] ? sync_one_sb+0x20/0x20
[    0.009999]  [<c1097e25>] kdb_bp+0x1d5/0x260
[    0.009999]  [<c111cd30>] ? sync_one_sb+0x20/0x20
[    0.009999]  [<c1094ecd>] kdb_parse+0x2ed/0x630
[    0.009999]  [<c1097950>] ? kdb_bt+0x3b0/0x3b0
[    0.009999]  [<c1913208>] ? kdb_initbptab+0x188/0x18e
[    0.009999]  [<c191301b>] kdb_init+0x5d2/0x637
[    0.009999]  [<c19129f1>] dbg_late_init+0x22/0x24
[    0.009999]  [<c18fd7a3>] start_kernel+0x2a8/0x2ea
[    0.009999]  [<c18fd325>] ? kernel_init+0x13a/0x13a
[    0.009999]  [<c18fd0b3>] i386_start_kernel+0xb3/0xbb

=============================
Tim Bird
Architecture Group Chair, CE Workgroup of the Linux Foundation Senior Staff Engineer, Sony Network Entertainment
=============================

maintainer
#3 lore raw
In reply to Tim Bird
Quoted 5 lines · Tim Bird

This fixes a bug with setting a breakpoint during kdb initialization (from kdb_cmds). Any call to kdb_printf() before the initialization of the kgdboc serial console driver (which happens much later during bootup than kdb_init), results in kernel panic due to the use of dbg_io_ops before it is initialized.

I added this patch to the merge queue for the 3.4 kernel.

In the test and review of this patch, I thought there might be another race condition with kdb pager, but this is not the case. The run time validation shows that the kdb pager can only get activated by entering and configuring the kdb shell and it is turned off on exit, so we are all set and no further changes are needed.

Thanks,
Jason.

author
#4 lore raw
Quoted 7 lines · Jason Wessel
Quoted 5 lines · Tim Bird

This fixes a bug with setting a breakpoint during kdb initialization (from kdb_cmds). Any call to kdb_printf() before the initialization of the kgdboc serial console driver (which happens much later during bootup than kdb_init), results in kernel panic due to the use of dbg_io_ops before it is initialized.

Quoted 7 lines

I added this patch to the merge queue for the 3.4 kernel.

In the test and review of this patch, I thought there might be another race condition with kdb pager, but this is not the case. The run time validation shows that the kdb pager can only get activated by entering and configuring the kdb shell and it is turned off on exit, so we are all set and no further changes are needed.

Thanks!
-- Tim

=============================
Tim Bird
Architecture Group Chair, CE Workgroup of the Linux Foundation Senior Staff Engineer, Sony Network Entertainment
=============================