1469 2011-09-10 09:32:56 +0000 qemuarm images have blue and red component channels reversed in X/pslash 2011-09-27 23:06:21 +0000 1 1 6 Yocto Project Subprojects Kernel kernel-tooling unspecified x86 arm VERIFIED FIXED High normal 1.1 1 richard.purdie bruce.ashfield dvhart liming.wang poky.bs.watcher poky.watcher yi.zhao yp.kernel.watcher yp.watcher --- --- --- oldest_to_newest 16128 0 richard.purdie 2011-09-10 09:32:56 +0000 If you look at X or at psplash the blue and red changes are reversed (the yocto dot on the logo is orange, not blue as it should be). This is likely a kernel issue reporting the wrong bitmasks to userspace since the pslash code hasn't changed recently (and for X and psplash to be showing the same problem). 16227 1 dvhart 2011-09-15 15:06:01 +0000 Bruce, can you ask someone to look into this? I wonder if this may also be related to Bug 1260. 16234 2 bruce.ashfield 2011-09-15 17:53:05 +0000 We can have a look. I've never noticed this myself (and have tested sato). The kernel has been stable for some time, so if this is kernel, it's been there since we jumped to 3.0. 16247 3 liming.wang 2011-09-16 03:07:25 +0000 9728c1b6a724daefc413b44e10253cdbb5e06d08 imports this issue. Specially, following line leads to the issue: + /* Always clear BGR for RGB565: we do the routing externally */ + if (fb->fb.var.green.length == 6) + regs->cntl &= ~CNTL_BGR; 16248 4 richard.purdie 2011-09-16 03:22:41 +0000 Ok, this looks like an intentional upstream change and it therefore looks like the emulator is now "wrong" compared to what the real harware would do. We likely therefore need to fix qemu to handle this situation correctly. The function in question in pl110_update_display(). 16250 5 238 richard.purdie 2011-09-16 06:46:36 +0000 Created attachment 238 Patch demonstrating fixed colours on qemuarm 16251 6 bruce.ashfield 2011-09-16 06:48:47 +0000 Who's running with getting the fix into the tree ? I just want to be sure. 16258 7 richard.purdie 2011-09-16 07:59:14 +0000 Since I was experimenting and had a reproducer I've put http://git.openembedded.org/cgit.cgi/openembedded-core-contrib/commit/?h=rpurdie/qemu&id=d06781eab16881d2206a3306c66a0c4c1e178dde together so I'll take it. 16259 8 bruce.ashfield 2011-09-16 08:03:41 +0000 Great. And the patch looks fine to my eye. 16266 9 richard.purdie 2011-09-16 09:25:18 +0000 Fixed in master: http://git.yoctoproject.org/cgit.cgi/poky/commit/?id=bb0cc3cfa69ce02a25e5b7b22df0e5554f6592fb 16540 10 yi.zhao 2011-09-27 23:06:21 +0000 Verified it with 1.1 M4 RC3 20110925 build. Tree/Branch: poky/edison Poky Commit: 41c564fe605fa4685fd27b1721e3424ac8a96db1 238 2011-09-16 06:46:36 +0000 2011-09-16 06:47:20 +0000 Patch demonstrating fixed colours on qemuarm bgr.patch text/plain 577 richard.purdie SW5kZXg6IHFlbXUtMC4xNC4wL2h3L3BsMTEwLmMKPT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PQotLS0gcWVtdS0wLjE0LjAu b3JpZy9ody9wbDExMC5jCTIwMTEtMDktMTYgMTQ6NDU6MzQuMjI4NjY4NTE0ICswMTAwCisrKyBx ZW11LTAuMTQuMC9ody9wbDExMC5jCTIwMTEtMDktMTYgMTQ6NDU6MzguMzU4NjY4NTIwICswMTAw CkBAIC0xNDEsMTAgKzE0MSwxMCBAQAogICAgICAgICBmcHJpbnRmKHN0ZGVyciwgInBsMTEwOiBC YWQgY29sb3IgZGVwdGhcbiIpOwogICAgICAgICBleGl0KDEpOwogICAgIH0KLSAgICBpZiAocy0+ Y3IgJiBQTDExMF9DUl9CR1IpCisgICAgLy9pZiAocy0+Y3IgJiBQTDExMF9DUl9CR1IpCiAgICAg ICAgIGJwcF9vZmZzZXQgPSAwOwotICAgIGVsc2UKLSAgICAgICAgYnBwX29mZnNldCA9IDE4Owor ICAgIC8vZWxzZQorICAgIC8vICAgIGJwcF9vZmZzZXQgPSAxODsKIAogICAgIGlmIChzLT5jciAm IFBMMTEwX0NSX0JFQk8pCiAgICAgICAgIGZuID0gZm50YWJsZVtzLT5icHAgKyA2ICsgYnBwX29m ZnNldF07Cg==