Latest Bugs & Tasks
Confirmed fixed, thanks!
I've looked into it and this is now a won't-fix for me. See hrev46903.
resolved as of hrev46902. Confirmed in llvm that they moved from using 1ull to UINT64_C(1) recently uncovering this issue.
LLVM 3.4 compiled successfully as did x86_64 Haiku.
Fixed in hrev46900.
With the new scheduler, I get very slow graphics drawing on my machine. Dragging windows around is slow and lagging, same with e.g. scrolling a website in WebPositive. Fluid video playback is impossible.
A good way to benchmark this is the "Chart" application from our Demos folder. I tested with 3 Haiku revisions:
- hrev46684 (nightly right before the scheduler merge)
- hrev46699 (nightly right after the scheduler merge)
- hrev46893 (latest nightly)
Selecting "DirectWindow" as display method in Chart, all three revisions get fluid 60fps with 3% CPU usage. However, using "DrawBitmap", the problem becomes apparent: only the pre-scheduler-merge version performs well and fluid, the others suffer from very bad performance (less than 10fps even at 100% CPU usage).
I'm attaching 3 screenshots of the test which also show ActivityMonitor.
Test system: AMD E-450 APU (dual core), Radeon HD6320, 2GB RAM. Same behaviour with both, radeon_hd and VESA graphics driver.
Let me know if further information is required.
This is a trick used in the Clover bootloader to get native video modes from VESA.
The idea is to patch the VESA BIOS (which is shadowed in RAM) and replace the video mode tables with the EDID data.
This gets native video mode on ATI, NVidia and Intel cards, without the need to write an actual driver. Their Vesa BIOS is smart enough to accept any resolution in the table.
After creating the "Intel extended partition" can not create any partition.
The window is clean.
LLVM doesn't build on x86_64. Tracked it down to a bug in our stdint.h. I found the solution when I ran onto a DragonflyBSD post seeing the same LLVM build error.
That shows a fix for the same issue present in our stdint.h
Working on a patch now, but need to do some in-depth testing before I commit it.
I'm having an intermittent problem with crashes at bootup that I really need to solve, and I'm offering a $200 bounty in case someone can solve this for me. Here's the scenario.
I can boot OK an average of about 70 percent of the time, maybe a bit more. But sometimes, just after the desktop and icons appears, I see a row of gold sparkles across the top of the screen for about one second (about 30 pixels tall) and then Haiku KDLs. I have attached a screen grab and can get more of them if it's helpful.
I'm using HREV 46104, one of the last builds before PM, because there are still a few rough edges that keep us from being able to move all our systems builds over to PM. The motherboard is an ASUS F2A55M and it uses an AMD processor. I have tried both IDE emulation and AHCI SATA modes in BIOS and the problem happens with both.
Would be most grateful to have this solved!
GitHub releases expect you to attach downloads by drag and drop. However, Web+ intercepts drop of a file and loads it, so this doesn't work.
hrev46685_gcc2_hybrid - anyboot image from USB drive
ATI Xpress 1150 on board graphics - black screen after booting
Radeon HD2400 - white screen after booting
Radeon HD3450 - white screen after booting
Radeon HD3470 - white screen after booting
Attached serial debug logs for each graphics chipset.
Haiku alpha4.1 confirmed working at native resolution(1920x1080)
of monitor, using radeon driver, with all graphics chipsets except
ATI Xpress 1150 on-board.
Thanks for the heads up!
Steps to reproduce:
- Login to a Wordpress site using an Admin user
- Notice that it logs in without issue
- Click on any WP menu (like Dashboard, etc.)
- Notice that it redirects you to the login screen
Here is the actual URL it redirects to (on my WP site):
Looks like it's try to 're-authorize', for some reason...Although, it never takes you to the WP dashboard, etc.