| Message ID | 1581066052-25269-1-git-send-email-vrd@amazon.de (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | kimage: Allow load/exec for images with KEXEC_TYPE_LIVE_UPDATE | expand |
On 07.02.2020 10:00, Varad Gautam wrote: > Do not -EINVAL on loading/execing an image if kexec type is > KEXEC_TYPE_LIVE_UPDATE. > > Signed-off-by: Varad Gautam <vrd@amazon.de> > CC: David Woodhouse <dwmw@amazon.co.uk> Please Cc maintainers on patches. > --- a/xen/common/kimage.c > +++ b/xen/common/kimage.c > @@ -421,6 +421,7 @@ struct page_info *kimage_alloc_control_page(struct kexec_image *image, > switch ( image->type ) > { > case KEXEC_TYPE_DEFAULT: > + case KEXEC_TYPE_LIVE_UPDATE: > pages = kimage_alloc_normal_control_page(image, memflags); > break; > case KEXEC_TYPE_CRASH: > @@ -775,6 +776,7 @@ static int kimage_load_segment(struct kexec_image *image, xen_kexec_segment_t *s > switch ( image->type ) > { > case KEXEC_TYPE_DEFAULT: > + case KEXEC_TYPE_LIVE_UPDATE: > result = kimage_load_normal_segment(image, segment); > break; > case KEXEC_TYPE_CRASH: > @@ -803,6 +805,7 @@ int kimage_alloc(struct kexec_image **rimage, uint8_t type, uint16_t arch, > switch( type ) > { > case KEXEC_TYPE_DEFAULT: > + case KEXEC_TYPE_LIVE_UPDATE: > result = kimage_normal_alloc(rimage, entry_maddr, nr_segments, segment); > break; > case KEXEC_TYPE_CRASH: You don't mention any prereq patches, and I can't see any definition of KEXEC_TYPE_LIVE_UPDATE in the public headers. IOW I can't see how this patch would be able to not break the build on current staging. Please clarify. Jan
On Fri, 2020-02-07 at 10:24 +0100, Jan Beulich wrote: > You don't mention any prereq patches, and I can't see any > definition of KEXEC_TYPE_LIVE_UPDATE in the public headers. > IOW I can't see how this patch would be able to not break > the build on current staging. Please clarify. I don't think this was meant to go to the list. It's a fixup for the patch I posted in my RFC series at https://lists.xenproject.org/archives/html/xen-devel/2020-01/msg02690.html The original testing was just hard-coding addresses and using a 'normal' kexec load, and we were only testing KEXEC_TYPE_LIVE_UPDATE for the actual kexec_exec. Once Varad had implemented the automatic placement support in kexec userspace tools, we spotted that the load operation wasn't working for KEXEC_TYPE_LIVE_UPDATE, hence this fixup. I merged it into the appropriate patch in my tree some time last week. https://xenbits.xen.org/gitweb/?p=people/dwmw2/xen.git;a=commitdiff;h=0bf4204cdb5f4fb8d17ead8df3e4fda37cfd6a28
diff --git a/xen/common/kimage.c b/xen/common/kimage.c index 86d2797..599aa74 100644 --- a/xen/common/kimage.c +++ b/xen/common/kimage.c @@ -421,6 +421,7 @@ struct page_info *kimage_alloc_control_page(struct kexec_image *image, switch ( image->type ) { case KEXEC_TYPE_DEFAULT: + case KEXEC_TYPE_LIVE_UPDATE: pages = kimage_alloc_normal_control_page(image, memflags); break; case KEXEC_TYPE_CRASH: @@ -775,6 +776,7 @@ static int kimage_load_segment(struct kexec_image *image, xen_kexec_segment_t *s switch ( image->type ) { case KEXEC_TYPE_DEFAULT: + case KEXEC_TYPE_LIVE_UPDATE: result = kimage_load_normal_segment(image, segment); break; case KEXEC_TYPE_CRASH: @@ -803,6 +805,7 @@ int kimage_alloc(struct kexec_image **rimage, uint8_t type, uint16_t arch, switch( type ) { case KEXEC_TYPE_DEFAULT: + case KEXEC_TYPE_LIVE_UPDATE: result = kimage_normal_alloc(rimage, entry_maddr, nr_segments, segment); break; case KEXEC_TYPE_CRASH:
Do not -EINVAL on loading/execing an image if kexec type is KEXEC_TYPE_LIVE_UPDATE. Signed-off-by: Varad Gautam <vrd@amazon.de> CC: David Woodhouse <dwmw@amazon.co.uk> --- xen/common/kimage.c | 3 +++ 1 file changed, 3 insertions(+)