* [PATCH] runqemu: support uImage kernel @ 2021-04-16 16:43 Adrian Freihofer 2021-04-16 18:06 ` [OE-core] " Bruce Ashfield 0 siblings, 1 reply; 3+ messages in thread From: Adrian Freihofer @ 2021-04-16 16:43 UTC (permalink / raw) To: openembedded-core; +Cc: Adrian Freihofer u-boot's uImage kernel format is not supported by Qemu. Extract the kernel from the uImage to a temporary directory. Signed-off-by: Adrian Freihofer <adrian.freihofer@siemens.com> --- scripts/runqemu | 24 ++++++++++++++++++++++-- 1 file changed, 22 insertions(+), 2 deletions(-) diff --git a/scripts/runqemu b/scripts/runqemu index ba0b701aff..fa630fc7a4 100755 --- a/scripts/runqemu +++ b/scripts/runqemu @@ -163,6 +163,7 @@ class BaseConfig(object): self.bios = '' self.kernel_cmdline = '' self.kernel_cmdline_script = '' + self.temp_kernel_dir = None self.bootparams = '' self.dtb = '' self.fstype = '' @@ -1422,10 +1423,26 @@ class BaseConfig(object): self.setup_serial() self.setup_vga() + def dump_uimage(self): + if self.kernel and self.get('KERNEL_IMAGETYPE') == 'uImage': + import tempfile + self.temp_kernel_dir = tempfile.mkdtemp() + kernel = os.path.join(self.temp_kernel_dir, 'image') + dumpimage_bin = os.path.join(self.bindir_native, 'dumpimage') + image_link_name = self.get('IMAGE_LINK_NAME') + logger.info('Running: %s' % str([dumpimage_bin, "-o ", kernel, self.kernel])) + staging_bindir_native = '%s/usr/bin' % self.get('STAGING_DIR_NATIVE') + dumpimage_bin = os.path.join(staging_bindir_native, 'dumpimage') + subprocess.run([dumpimage_bin, "-p", "0", "-o", kernel, self.kernel]) + return kernel + return self.kernel + def start_qemu(self): import shlex - if self.kernel: - kernel_opts = "-kernel %s -append '%s %s %s %s'" % (self.kernel, self.kernel_cmdline, + + kernel = self.dump_uimage() + if kernel: + kernel_opts = "-kernel %s -append '%s %s %s %s'" % (kernel, self.kernel_cmdline, self.kernel_cmdline_script, self.get('QB_KERNEL_CMDLINE_APPEND'), self.bootparams) if self.bios: @@ -1481,6 +1498,9 @@ class BaseConfig(object): shutil.rmtree(self.rootfs) shutil.rmtree('%s.pseudo_state' % self.rootfs) + if self.temp_kernel_dir: + shutil.rmtree(self.temp_kernel_dir) + self.cleaned = True def run_bitbake_env(self, mach=None): -- 2.26.3 ^ permalink raw reply related [flat|nested] 3+ messages in thread
* Re: [OE-core] [PATCH] runqemu: support uImage kernel 2021-04-16 16:43 [PATCH] runqemu: support uImage kernel Adrian Freihofer @ 2021-04-16 18:06 ` Bruce Ashfield 2021-04-18 13:53 ` Adrian Freihofer 0 siblings, 1 reply; 3+ messages in thread From: Bruce Ashfield @ 2021-04-16 18:06 UTC (permalink / raw) To: Adrian Freihofer Cc: Patches and discussions about the oe-core layer, Adrian Freihofer On Fri, Apr 16, 2021 at 12:43 PM Adrian Freihofer <adrian.freihofer@gmail.com> wrote: > > u-boot's uImage kernel format is not supported by Qemu. Extract the > kernel from the uImage to a temporary directory. > I'd suggest that we should document in the commit log why someone wouldn't just build a kernel image format that is already supported by runqemu (which is what I typically do). I'm guessing it is some sort of h/w boot being re-purposed to qemu emulation, and this is to avoid building a second kernel image ? Additionally, this should have some sort of test, or the code is going to bitrot due to not being run by anything in core. > Signed-off-by: Adrian Freihofer <adrian.freihofer@siemens.com> > --- > scripts/runqemu | 24 ++++++++++++++++++++++-- > 1 file changed, 22 insertions(+), 2 deletions(-) > > diff --git a/scripts/runqemu b/scripts/runqemu > index ba0b701aff..fa630fc7a4 100755 > --- a/scripts/runqemu > +++ b/scripts/runqemu > @@ -163,6 +163,7 @@ class BaseConfig(object): > self.bios = '' > self.kernel_cmdline = '' > self.kernel_cmdline_script = '' > + self.temp_kernel_dir = None > self.bootparams = '' > self.dtb = '' > self.fstype = '' > @@ -1422,10 +1423,26 @@ class BaseConfig(object): > self.setup_serial() > self.setup_vga() > > + def dump_uimage(self): > + if self.kernel and self.get('KERNEL_IMAGETYPE') == 'uImage': > + import tempfile > + self.temp_kernel_dir = tempfile.mkdtemp() > + kernel = os.path.join(self.temp_kernel_dir, 'image') > + dumpimage_bin = os.path.join(self.bindir_native, 'dumpimage') > + image_link_name = self.get('IMAGE_LINK_NAME') > + logger.info('Running: %s' % str([dumpimage_bin, "-o ", kernel, self.kernel])) > + staging_bindir_native = '%s/usr/bin' % self.get('STAGING_DIR_NATIVE') > + dumpimage_bin = os.path.join(staging_bindir_native, 'dumpimage') > + subprocess.run([dumpimage_bin, "-p", "0", "-o", kernel, self.kernel]) > + return kernel > + return self.kernel > + > def start_qemu(self): > import shlex > - if self.kernel: > - kernel_opts = "-kernel %s -append '%s %s %s %s'" % (self.kernel, self.kernel_cmdline, > + > + kernel = self.dump_uimage() > + if kernel: > + kernel_opts = "-kernel %s -append '%s %s %s %s'" % (kernel, self.kernel_cmdline, > self.kernel_cmdline_script, self.get('QB_KERNEL_CMDLINE_APPEND'), > self.bootparams) If the kernel never fails from the dup_uimage() command, why test for it ? I don't see how the routine could return something that doesn't pass the test. If so, aren't we going to run into more cryptic errors (or traceback) later if there was some sort of failure. The existing code doesn't look like it is completely error proofed either, that just came to mind when looking at the patch. Bruce > if self.bios: > @@ -1481,6 +1498,9 @@ class BaseConfig(object): > shutil.rmtree(self.rootfs) > shutil.rmtree('%s.pseudo_state' % self.rootfs) > > + if self.temp_kernel_dir: > + shutil.rmtree(self.temp_kernel_dir) > + > self.cleaned = True > > def run_bitbake_env(self, mach=None): > -- > 2.26.3 > > > > -- - Thou shalt not follow the NULL pointer, for chaos and madness await thee at its end - "Use the force Harry" - Gandalf, Star Trek II ^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: [OE-core] [PATCH] runqemu: support uImage kernel 2021-04-16 18:06 ` [OE-core] " Bruce Ashfield @ 2021-04-18 13:53 ` Adrian Freihofer 0 siblings, 0 replies; 3+ messages in thread From: Adrian Freihofer @ 2021-04-18 13:53 UTC (permalink / raw) To: Bruce Ashfield; +Cc: Patches and discussions about the oe-core layer Hi Bruce, Thank you for asking the right questions. This patch should be ignored. On Fri, 2021-04-16 at 14:06 -0400, Bruce Ashfield wrote: > On Fri, Apr 16, 2021 at 12:43 PM Adrian Freihofer > <adrian.freihofer@gmail.com> wrote: > > u-boot's uImage kernel format is not supported by Qemu. Extract the > > kernel from the uImage to a temporary directory. > > > > I'd suggest that we should document in the commit log why someone > wouldn't just build a kernel image format that is already supported by > runqemu (which is what I typically do). I'm guessing it is some sort > of h/w boot being re-purposed to qemu emulation, and this is to avoid > building a second kernel image ? > The runqemu script explicitly checks for the uImage type. However, my conclusion so far is that Qemu does not support this format, or at least it is broken. I've been thinking about adding more reasonable feedback to the user instead of just launching the uImage and running into an infinite loop of nothing. However this would need deeper understanding of Qemus's support for e.g. uImage format or better ctest cases covering this use cases. > Additionally, this should have some sort of test, or the code is going > to bitrot due to not being run by anything in core. The current runquem-related test cases only cover the x86_64 MACHINE.Regardless of the discussion around this patch, it might be useful to cover more machines there as well. This would make it possible to test some more image formats, such as uImage and fitImage, which are more widely used on ARM machines. I will have a look at that. Regards, Adrian > > > Signed-off-by: Adrian Freihofer <adrian.freihofer@siemens.com> > > --- > > scripts/runqemu | 24 ++++++++++++++++++++++-- > > 1 file changed, 22 insertions(+), 2 deletions(-) > > > > diff --git a/scripts/runqemu b/scripts/runqemu > > index ba0b701aff..fa630fc7a4 100755 > > --- a/scripts/runqemu > > +++ b/scripts/runqemu > > @@ -163,6 +163,7 @@ class BaseConfig(object): > > self.bios = '' > > self.kernel_cmdline = '' > > self.kernel_cmdline_script = '' > > + self.temp_kernel_dir = None > > self.bootparams = '' > > self.dtb = '' > > self.fstype = '' > > @@ -1422,10 +1423,26 @@ class BaseConfig(object): > > self.setup_serial() > > self.setup_vga() > > > > + def dump_uimage(self): > > + if self.kernel and self.get('KERNEL_IMAGETYPE') == 'uImage': > > + import tempfile > > + self.temp_kernel_dir = tempfile.mkdtemp() > > + kernel = os.path.join(self.temp_kernel_dir, 'image') > > + dumpimage_bin = os.path.join(self.bindir_native, 'dumpimage') > > + image_link_name = self.get('IMAGE_LINK_NAME') > > + logger.info('Running: %s' % str([dumpimage_bin, "-o ", kernel, self.kernel])) > > + staging_bindir_native = '%s/usr/bin' % self.get('STAGING_DIR_NATIVE') > > + dumpimage_bin = os.path.join(staging_bindir_native, 'dumpimage') > > + subprocess.run([dumpimage_bin, "-p", "0", "-o", kernel, self.kernel]) > > + return kernel > > + return self.kernel > > + > > def start_qemu(self): > > import shlex > > - if self.kernel: > > - kernel_opts = "-kernel %s -append '%s %s %s %s'" % (self.kernel, self.kernel_cmdline, > > + > > + kernel = self.dump_uimage() > > + if kernel: > > + kernel_opts = "-kernel %s -append '%s %s %s %s'" % (kernel, self.kernel_cmdline, > > self.kernel_cmdline_script, self.get('QB_KERNEL_CMDLINE_APPEND'), > > self.bootparams) > > If the kernel never fails from the dup_uimage() command, why test for > it ? I don't see how the routine could return something that doesn't > pass the test. If so, aren't we going to run into more cryptic errors > (or traceback) later if there was some sort of failure. > > The existing code doesn't look like it is completely error proofed > either, that just came to mind when looking at the patch. > > Bruce > > > if self.bios: > > @@ -1481,6 +1498,9 @@ class BaseConfig(object): > > shutil.rmtree(self.rootfs) > > shutil.rmtree('%s.pseudo_state' % self.rootfs) > > > > + if self.temp_kernel_dir: > > + shutil.rmtree(self.temp_kernel_dir) > > + > > self.cleaned = True > > > > def run_bitbake_env(self, mach=None): > > -- > > 2.26.3 > > > > > > > > > > ^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2021-04-18 13:53 UTC | newest] Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed) -- links below jump to the message on this page -- 2021-04-16 16:43 [PATCH] runqemu: support uImage kernel Adrian Freihofer 2021-04-16 18:06 ` [OE-core] " Bruce Ashfield 2021-04-18 13:53 ` Adrian Freihofer
This is an external index of several public inboxes, see mirroring instructions on how to clone and mirror all data and code used by this external index.