Opened 11 years ago

Closed 10 years ago

#2275 closed bug (fixed)

Radeon driver GCC4 issues.

Reported by: euan Owned by: euan
Priority: normal Milestone: R1
Component: Drivers/Graphics/radeon Version: R1/pre-alpha1
Keywords: Cc:
Blocked By: Blocking:
Has a Patch: no Platform: x86

Description

Radeon driver fails to run when built with GCC4. Bad stuff happens, and darkness is final outcome.

Log to follow.

Attachments (5)

radeon.serial.log (83.0 KB) - added by anevilyak 11 years ago.
Serial log from radeon error
syslog (308.0 KB) - added by anevilyak 11 years ago.
GCC2 Log for comparison
serial.gcc4.0529.log (75.3 KB) - added by anevilyak 11 years ago.
Updated log with higher Radeon debugging on.
radeon.gcc4.i2ctrace.log (77.3 KB) - added by anevilyak 11 years ago.
Log with I2C tracing enabled
radeon.gcc4.i2c-spin.debug.log (150.5 KB) - added by anevilyak 11 years ago.
Next log with debug tracing added to i2c.c's spin() implementation to ensure gcc4 wasn't optimizing it out somehow,

Download all attachments as: .zip

Change History (9)

Changed 11 years ago by anevilyak

Attachment: radeon.serial.log added

Serial log from radeon error

comment:1 Changed 11 years ago by anevilyak

FYI, the system itself appears to be running OK...keyboard is responding to numlock, I can ping, etc., it's just the display that's in an invalid mode.

Let me know what else I can do to help debug.

Changed 11 years ago by anevilyak

Attachment: syslog added

GCC2 Log for comparison

Changed 11 years ago by anevilyak

Attachment: serial.gcc4.0529.log added

Updated log with higher Radeon debugging on.

Changed 11 years ago by anevilyak

Attachment: radeon.gcc4.i2ctrace.log added

Log with I2C tracing enabled

Changed 11 years ago by anevilyak

Next log with debug tracing added to i2c.c's spin() implementation to ensure gcc4 wasn't optimizing it out somehow,

comment:2 Changed 11 years ago by anevilyak

Based on banging on things all weekend, I think it's possible we might be looking at the wrong problem...one of the recent changes in the tree caused whatever mode EDID detected during boot to be written to the app_server_settings file immediately to prevent a mode switch, which in turn short circuits the radeon's monitor detection routines...however, I still end up with the same invalid signal to the monitor. At this point I'm starting to think something's going wrong with the kernel driver's setup of things, since I also tried forcing the routines that query the BIOS for the connector types to force DVI, and that also made no difference.

comment:3 Changed 11 years ago by euan

Status: newassigned

It must be an i2c issue IMHO. it works in GCC2 and fails in gcc4.

comment:4 Changed 10 years ago by mmlr

Resolution: fixed
Status: assignedclosed

Fixed in hrev32593.

Note: See TracTickets for help on using tickets.