♟ scottl

In D28821#647027, @markj wrote:

In D28821#647026, @scottl wrote:

I won't be able to go back and test this for at least a few more days. However, when I tested the atomic_load_int() suggestion over the weekend, I got a really weird panic instead. I didn't dig into it because of other time pressures, but it does make me worry about compiler bugs and other weirdness, whereas the membar is proven to work for now.

Looking at the patch you pasted at the time, you had if (atomic_load_int(tracker->rmp_flags)) instead of if (atomic_load_int(&tracker->rmp_flags)). I'm not sure how the former would compile (though it is more plausible on armv7), but if it did I'd indeed expect a panic.

I won't be able to go back and test this for at least a few more days. However, when I tested the atomic_load_int() suggestion over the weekend, I got a really weird panic instead. I didn't dig into it because of other time pressures, but it does make me worry about compiler bugs and other weirdness, whereas the membar is proven to work for now.

Thank you for doing this. My main suggestion is to fix mps_read_extended_config_page so it returns IOCStatus already byte-swapped, which will eliminate a lot of the other diffs. I also recommend looking at mps_cmd.c and fixing code that is under #ifdef USE_MPT_IOCTLS as well as not

I'm good with this, actually thought that Alexander had already done it. This should go into a release not in case there are people maintaining out-of-tree drivers.

ACK, I'll look at it over the weekend.

Add myself to the handbook. Woohoo!

Add an email address for Adam Weinberger and remove some text that

Add the September-October 2002 Developers Status Report.

Flesh out the section on device hints a bit more and add explicit text

Update the webpage to match the revide schedule. Add some extra items to

Remove a bogud extra field from the last row.

Note that the tree has been branched and HEAD unfrozen. Also fix a small

amd(4) no longer uses vtophys().

Update the 5.0-RC2 TODO list.

Remove the item about converting the Perl scripts in the base system since

Note the addition of RC3 and adjust the dates for 5.0-R

Move the i386 bootblock item to RC3. Remove the item about OpenSSH since

Remove the TODO item for boot2, thanks to imp.

Remove the rpc socket rename item since it's complete.

The ACPI documentation item is a done as it's going to be.

Note that the ports tree was tagged this morning. Thanks Kris!

SHorten the 5.0 TODO list some more thanks to phk fixing boot0cfg

Add an item for starting the RC3 builds, and an item noting the desire to

Mark the sysinstall documentation and malloc items off the TODO list.

The sysinstall rpcbind issue has been committed. Remove it from the list.

The devd item is complete.

The devfs sweep is as done as it's going to be for now.

Move the remaining RC3 items to 5.0R and update their status.

Update the schedule with the actual RC3 start time and the added item

Update the schedule to note the release of RC3 and the new date to

Remove the TrustedBSD documentation, sysinstall, and disklabel issues as

Mark a few more completed items off.

Add the November-December Status Report.

Add entry for 5.0-RELEASE and note that 5-CURRENT is now as 500100

Remove a duplicate entry for 5.0-RELEASE

Forgot to ad my entry into the chapter

Add my key for fun and profit

Expand the busdma project page to also track SMPng lockdown and INTR_MPSAFE

Make a couple of small updates to the SMPng page

Add a link to the busdma/SMPng driver conversion webpage.

Add the 5-STABLE roadmap document. I do not hold an exclusive lock on

Link in the 5-STABLE roadmap document.

Note that the aac driver is INTR_MPSAFE capable and no longer unsafe for

Update that aac is now locked.

Update to note the __FreeBSD_version bump for taskqueue_swi.

Add the Jan-Feb 2003 FreeBSD status report.

Fix a small grammar nit.

Start the schedule and todo pages for 5.1.

Add a link to the 5.1 schedule page.

Hook the 5.1R directory up to the build.

Note that the aac driver is 64-bit capable

Add a reminder to turn off debugging for RELENG_5_1

Move if_wi, pae, and lazy context switch tasks from 'Must Have' to 'Must

Move the iir item from 'must fix' to 'must test'.

Note that the ACPI refresh is done and should be tested.

GBDE documentation is done, move 16byte CDB to the testing section, remove

Remove the 16-byte CDB as Ken feels that it has had sufficient testing.

Remove the SMP double fault item as it was most likely due to bad hardware.

Note that HEAD is now frozen.

Remove the 'no console when smbus enabled' testing item as the original

The compiler problems with XFree86-4-libraries on sparc64 have been worked

Move the sio/ia64 issue to the desired list and reword slightly.

Update the schedule to reflect that we are going to do BETA2 today

Move several items from the 5.1 list to the 5.2 list, update status on others

Update the schedule to reflect slipping the branch by three days. We are

Move the MAC frame work and rpc.lockd items to the 5.2 list. NFS locking

Move the truss race item to 5.2 since it won't get done for 5.1. Move the

The busdma manpage has been committed, remove it from the list. Also

Check off the branch and unfreeze boxes

ipfw2 appears to work on strict-align platforms. Thanks to Bernd and

Bring the schedule up to date.

Slightly update branch status

Update porters handbook for the 501102 busdma API change.

Take GCC 3.3 off the list since it's done. Thanks to Alexander and everyone

RELENG_5_1 has been handed off to the Security Officer team.

Update the 5.2 TODO list to reflect progress in many areas.

Update the 5-stable roadmap. Remove items that are done or no londer

Add two tables to show THR and KSE status. If anyone knows how to make the

Marcel has fixed KSE on ia64, mark it as working in the status table.

The KSE-ia64 and UART tasks are complete to satisfaction. Remove them from

Update the wording of the AMD64 task to match reality.

Add the March-September 2003 status report.

Hook the new status report into the website.

Update the 5.2-TODO list with a new section of 'showstopper defects'.

Add PR kern/58228 to the list of show-stopper bugs for 5.2.

Add an item about teaching sysinstall and libdisk about GDT partitions.

Re-arrange a few items at the request of their owners. Remove ia64

Introduce the schedule for the 5.2 release.

Document the PSE-related panics as a showstopper for 5.2.

Move the GPT sysinstall item to the 'please test' list. Thanks to Marcel

Move the ACPI-PCI task to the 'test' section. Thanks to John Baldwin for

scottl (Scott Long)
User

Projects

User Details

Recent Activity
View All

Wed, Feb 24

Tue, Feb 23

Sat, Feb 20

Tue, Feb 9

Tue, Feb 2

Dec 31 2020

Dec 24 2020

Dec 21 2020

Dec 13 2020

scottl (Scott Long)User