On Thu, Nov 21, 2024 at 12:37:41PM +0800, Philip Li wrote: > On Tue, Nov 19, 2024 at 09:19:25AM -0600, Bjorn Helgaas wrote: > > On Mon, Nov 18, 2024 at 12:01:12AM +0800, kernel test robot wrote: > > > tree/branch: https://git.kernel.org/pub/scm/linux/kernel/git/pci/pci.git controller/rockchip > > > branch HEAD: 592aac418ebdf451fe9b146bc2ca6dfc96921af0 PCI: rockchip-ep: Handle PERST# signal in endpoint mode > > > > > x86_64 allyesconfig clang-19 > > Let me double check this why it is success. Sorry about the confusing PASS info here, i have confirmed the error is there for head 592aac41. Below is the explanation Early on around "Date: Thu, 14 Nov 2024 17:57:46 +0800", the bot sends out mail "[pci:controller/rockchip] BUILD REGRESSION 337657a3c24c92befb3ed11d6f15402faa09f7dd" with 2 bisected reports https://lore.kernel.org/oe-kbuild-all/202411141106.4hI5VqIa-lkp@xxxxxxxxx https://lore.kernel.org/oe-kbuild-all/202411141621.uwFAKZb2-lkp@xxxxxxxxx Later for the new series with head 592aac41, it considers the bug is already there (reported before), thus is not newly introduced by 592aac41. So it marks build success. This tries to mean that the success is no new issues are introduced, though there's build failure when building the head. For example, when a branch is based on another branch, that base branch could contain build failure, but the newly added commits doesn't add new failures/warnings. Under this situation, the bot sends out "BUILD SUCCESS" mail. Hope this is helpful, and let me think of whether we can expose more info for situation like this to avoid confusion. Thanks > > Actually, the bot report "implicit declaration of function 'irq_set_status_flags'" at [1] > with sparc64-linux-gcc and sparc-allmodconfig, when the head is 337657a3c24c92befb3ed11d6f15402faa09f7dd. > > [1] https://lore.kernel.org/oe-kbuild-all/202411141621.uwFAKZb2-lkp@xxxxxxxxx/ > > > > > How can I reproduce this build? Do you have a packaged clang-19 > > toolchain? > > The clang package can be found at https://cdn.kernel.org/pub/tools/llvm/files. > > > > > The x86_64 allyesconfig build succeeded for the robot, but when I > > build on x86_64 with gcc-11.4.0, I get an error: > > > > $ gcc -v > > gcc version 11.4.0 (Ubuntu 11.4.0-1ubuntu1~22.04) > > $ git checkout 592aac418ebd > > $ make allyesconfig > > $ make drivers/pci/controller/pcie-rockchip-ep.o > > CC drivers/pci/controller/pcie-rockchip-ep.o > > drivers/pci/controller/pcie-rockchip-ep.c:640:9: error: implicit declaration of function ‘irq_set_irq_type’ > > > > irq_set_irq_type() is declared in <linux/irq.h>. On arm64, where this > > driver is used, <linux/irq.h> is included via this path: > > > > linux/pci.h > > linux/interrupt.h > > linux/hardirq.h > > arch/arm64/include/asm/hardirq.h > > asm-generic/hardirq.h > > linux/irq.h > > > > but on x86, arch/x86/include/asm/hardirq.h does not include > > asm-generic/hardirq.h and therefore doesn't include <linux/irq.h>. > > > > I'm confused about why the robot reported a successful build with > > clang-19. It seems like that should have the same problem I saw with > > gcc, so I'd like to try it manually. > > > > Bjorn > >
