MySQL Bugs: #109671: Crash after a select

Bug #109671	Crash after a select
Submitted:	18 Jan 9:08	Modified:	22 Feb 18:59
Reporter:	Paolo Meroni	Email Updates:
Status:	Verified	Impact on me:	None
Category:	MySQL Workbench	Severity:	S1 (Critical)
Version:	8.0.32	OS:	MacOS (Ventura)
Assigned to:		CPU Architecture:	Any (M1, x86)
Tags:	crash

Description:
Doing any select on a mysql 5.7 db will crash the program.
I can't check if the problem persists with different mysql version.

Starting mysqlworkbench from command line gives the following message after the crash:
2023-01-18 09:52:47.796 MySQLWorkbench[62617:4862748] invalid icon for toolbar
2023-01-18 09:52:47.895 MySQLWorkbench[62617:4862748] -[MQIndicatorCell setIsNull:]: unrecognized selector sent to instance 0x600000eb0870
zsh: illegal hardware instruction  ./MySQLWorkbench

How to repeat:
- Install mysqlworkbench 8.0.32 on a ARM M1 mac.
- Open a mysql 5.7 db.
- Do a select on any table

Crash report

Attachment: MySQLWorkbench-2023-01-18-095248.ips (application/octet-stream, text), 24.23 KiB.

I experience the same bug. For the avoidance of doubt, it worked fine with 8.0.31

Some versions and architecture (mysqlworkbench 8.0.32 on a ARM M1 mac).
Connecting to a different server version - in my case MySQL 8.0.16
Same method to precipitate a crash: so a select on any table (or any view, as it happens)

MacOS crash reporter says:

Exception Type:        EXC_BAD_INSTRUCTION (SIGILL)
Exception Codes:       0x0000000000000001, 0x0000000000000000

Termination Reason:    Namespace SIGNAL, Code 4 Illegal instruction: 4

Confirming that this occurs on MySQLWorkbench 8.0.32 on a ARM M1 Mac.

Opening an any table the running a select causes the crash.  8.0.31 doesn't have this issue.

/Applications/MySQLWorkbench.app/Contents/MacOS/MySQLWorkbench
2023-01-20 15:28:55.479 MySQLWorkbench[94775:9645584] -[MQIndicatorCell setIsNull:]: unrecognized selector sent to instance 0x60000272b1b0
zsh: illegal hardware instruction  /Applications/MySQLWorkbench.app/Contents/MacOS/MySQLWorkbench

Hello Paolo Meroni,

Thank you for the bug report. 
We are able to reproduce the issue on MacOS with the help of a colleague but we found a similar one i.e Bug #102284, please see Bug #102284.
Hence, If you have no objections then I would like to close this as a duplicate of Bug #102284. Thank you.

Regards,
Ashwini Patil

Hello Ashwini Patil,

This bug does *not* look like a dupe of Bug #102284 -- that bug:
 
* Crashes on startup rather than crashes on select.
* Is reported on Intel not Arm
* Affected 8.0.23 but not 8.0.22 -- this bug affects 8.0.32 but not 8.0.31

Thank you, Russ.

Regards,
Ashwini Patil

Bug #109686 marked as duplicate of this one

Bug #109729 marked as duplicate of this one

Bug #109732 marked as duplicate of this one

Bug #109734 marked as duplicate of this one

Bug #109766 marked as duplicate of this one

From 109732:

The problem is here in, lines 458 on in MResultsetViewer.mm.  It is assuming aCell is an MQResultSetCell that has setIsNull and setIsBlob whereas in fact aCell is an MQIndicatorCell which has neither.  Adding dummy setIsNull and setIsBlob methods to MQIndicatorCell prevents the crash.

Whether aCell being an MQIndicatorCell is expected or not is unknown to me and needs further investigation.

  if (columnIndex >= 0)
    {
      [aCell setIsNull: (*mData)->is_field_null(rowIndex, columnIndex)];
      [aCell setIsBlob: (*mData)->get_column_type(columnIndex) == bec::GridModel::BlobType];
    }
    else
    {
      [aCell setIsNull: NO];
      [aCell setIsBlob: NO];
    }

Bug #109783 marked as duplicate of this one

Bug #109778 marked as duplicate of this one

Bug #109801 marked as duplicate of this one

I can confirm this also happens on an Intel Mac (2.4 Ghz, 8 Core i9) running macOS 13.1. All forms of select as well as "show tables" immediately causes MySQL Workbench to crash.

Unfortunately, I'm experiencing the same crash while working with MySql Workbench. However, when I downloaded the older version (mysql-workbench-community-8.0.31-macos-x86_64.dmg), it worked perfectly fine. I'm using a MacBook pro M1 arm64.

IF you feel like building from source (clone from GitHub or you'll be in a world of hurt), the following change will prevent the crash:

diff --git a/frontend/mac/resultset/MQIndicatorCell.h b/frontend/mac/resultset/MQIndicatorCell.h
index c0f2ba60f..6832ac2f0 100644
--- a/frontend/mac/resultset/MQIndicatorCell.h
+++ b/frontend/mac/resultset/MQIndicatorCell.h
@@ -33,4 +33,7 @@
 - (void)setSelected:(BOOL)flag;
 - (void)setPlaceholder:(BOOL)flag;
 
+- (void)setIsNull:(BOOL)flag;
+- (void)setIsBlob:(BOOL)flag;
+
 @end
diff --git a/frontend/mac/resultset/MQIndicatorCell.m b/frontend/mac/resultset/MQIndicatorCell.m
index 2593016b8..de7295936 100644
--- a/frontend/mac/resultset/MQIndicatorCell.m
+++ b/frontend/mac/resultset/MQIndicatorCell.m
@@ -102,6 +102,17 @@ - (void)setPlaceholder: (BOOL)flag {
   _placeholder= flag;
 }
 
+//----------------------------------------------------------------------------------------------------------------------
+
+- (void)setIsBlob:(BOOL)flag {
+}
+
+//----------------------------------------------------------------------------------------------------------------------
+
+- (void)setIsNull:(BOOL)flag {
+}
+
+
 @end
 
 //----------------------------------------------------------------------------------------------------------------------

Bug #109849 marked as duplicate of this one

Bug #109855 marked as duplicate of this one

Bug #109930 marked as duplicate of this one

Bug #109924 marked as duplicate of this one

Bug #109925 marked as duplicate of this one

Bug #109932 marked as duplicate of this one

Bug #109927 marked as duplicate of this one

Bug #109933 marked as duplicate of this one

Bug #109995 marked as duplicate of this one

Bug #110035 marked as duplicate of this one

Bug #110143 marked as duplicate of this one

Hi you could change the CPU Architecture on the bug summary header to include both Intel and ARM.

added x86 architecture among CPUs

Bug #110173 marked as duplicate of this one

Bug #110240 marked as duplicate of this one

Bug #110248 marked as duplicate of this one

Bug #110249 marked as duplicate of this one

Bug #110264 marked as duplicate of this one

Bug #110308 marked as duplicate of this one

Bug #110324 marked as duplicate of this one

Bug #110351 marked as duplicate of this one

Error on my MAC

Attachment: errormysqlwb.txt (text/plain), 42.52 KiB.

Bug #110403 marked as duplicate of this one

Bug #110425 marked as duplicate of this one

Bug #110426 marked as duplicate of this one

Bug #110428 marked as duplicate of this one

Bug #110449 marked as duplicate of this one

Hi, are we getting closer to resolution, Oracle seems to be wasting more time closing other bug than resolving it, the solution is known. Thanks

Bug #110504 marked as duplicate of this one

Bug #110505 marked as duplicate of this one

Bug #110540 marked as duplicate of this one