| Message ID | 20241006214956.24339-9-dpsmith@apertussolutions.com (mailing list archive) |
|---|---|
| State | Superseded |
| Headers | show |
| Series | Boot modules for Hyperlaunch | expand |
On 2024-10-06 17:49, Daniel P. Smith wrote: > To allow a slow conversion of x86 over to struct boot_module, start with > replacing all references to struct mod to the early_mod element of struct > boot_module. These serves twofold, first to allow the incremental transition > from struct mod fields to struct boot_module fields. The second is to allow > the conversion of function definitions from taking struct mod parameters to > accepting struct boot_module as needed when a transitioned field will be > accessed. > > Signed-off-by: Daniel P. Smith <dpsmith@apertussolutions.com> > --- > xen/arch/x86/setup.c | 61 ++++++++++++++++++++++++-------------------- > 1 file changed, 34 insertions(+), 27 deletions(-) > > diff --git a/xen/arch/x86/setup.c b/xen/arch/x86/setup.c > index dd82ca3d43e2..ba4bee6b93af 100644 > --- a/xen/arch/x86/setup.c > +++ b/xen/arch/x86/setup.c > @@ -1341,15 +1341,15 @@ void asmlinkage __init noreturn __start_xen(unsigned long mbi_p) > set_kexec_crash_area_size((u64)nr_pages << PAGE_SHIFT); > kexec_reserve_area(); > > - initial_images = mod; > + initial_images = bi->mods[0].mod; Isn't this wrong? mod is the array of module_t * of *all* modules, but bi->mods[0].mod is a single module_t *? > > for ( i = 0; !efi_enabled(EFI_LOADER) && i < bi->nr_modules; i++ ) > { > - if ( mod[i].mod_start & (PAGE_SIZE - 1) ) > + if ( bi->mods[i].mod->mod_start & (PAGE_SIZE - 1) ) > panic("Bootloader didn't honor module alignment request\n"); > - mod[i].mod_end -= mod[i].mod_start; > - mod[i].mod_start >>= PAGE_SHIFT; > - mod[i].reserved = 0; > + bi->mods[i].mod->mod_end -= bi->mods[i].mod->mod_start; > + bi->mods[i].mod->mod_start >>= PAGE_SHIFT; > + bi->mods[i].mod->reserved = 0; > } > > /* > @@ -1509,13 +1510,15 @@ void asmlinkage __init noreturn __start_xen(unsigned long mbi_p) > #endif > } > > - if ( bi->mods[0].headroom && !mod->reserved ) > + if ( bi->mods[0].headroom && !bi->mods[0].mod->reserved ) > panic("Not enough memory to relocate the dom0 kernel image\n"); > for ( i = 0; i < bi->nr_modules; ++i ) > { > - uint64_t s = (uint64_t)mod[i].mod_start << PAGE_SHIFT; > + uint64_t s = (uint64_t)bi->mods[i].mod->mod_start > + << PAGE_SHIFT; pfn_to_paddr() ? > > - reserve_e820_ram(&boot_e820, s, s + PAGE_ALIGN(mod[i].mod_end)); > + reserve_e820_ram(&boot_e820, s, > + s + PAGE_ALIGN(bi->mods[i].mod->mod_end)); > } > > if ( !xen_phys_start ) > @@ -1593,8 +1596,9 @@ void asmlinkage __init noreturn __start_xen(unsigned long mbi_p) > map_e = boot_e820.map[j].addr + boot_e820.map[j].size; > for ( j = 0; j < bi->nr_modules; ++j ) > { > - uint64_t end = pfn_to_paddr(mod[j].mod_start) + > - mod[j].mod_end; > + uint64_t end = pfn_to_paddr( > + bi->mods[j].mod->mod_start) + > + bi->mods[j].mod->mod_end; I think you want a different indent. I think uint64_t end = pfn_to_paddr(bi->mods[j].mod->mod_start) will all fit on one line (indented all the way). (Thunderbird makes it difficult me to send indented.) > > if ( map_e < end ) > map_e = end; > @@ -1668,11 +1672,13 @@ void asmlinkage __init noreturn __start_xen(unsigned long mbi_p) > > for ( i = 0; i < bi->nr_modules; ++i ) > { > - set_pdx_range(mod[i].mod_start, > - mod[i].mod_start + PFN_UP(mod[i].mod_end)); > - map_pages_to_xen((unsigned long)mfn_to_virt(mod[i].mod_start), > - _mfn(mod[i].mod_start), > - PFN_UP(mod[i].mod_end), PAGE_HYPERVISOR); > + set_pdx_range(bi->mods[i].mod->mod_start, > + bi->mods[i].mod->mod_start + > + PFN_UP(bi->mods[i].mod->mod_end)); > + map_pages_to_xen( > + (unsigned long)mfn_to_virt(bi->mods[i].mod->mod_start), map_pages_to_xen((unsigned long)maddr_to_virt(bi->mods[i].start), All fits on one line. > + _mfn(bi->mods[i].mod->mod_start), > + PFN_UP(bi->mods[i].mod->mod_end), PAGE_HYPERVISOR); > } > > #ifdef CONFIG_KEXEC Regards, Jason
On 10/7/24 15:34, Jason Andryuk wrote: > On 2024-10-06 17:49, Daniel P. Smith wrote: >> To allow a slow conversion of x86 over to struct boot_module, start with >> replacing all references to struct mod to the early_mod element of struct >> boot_module. These serves twofold, first to allow the incremental >> transition >> from struct mod fields to struct boot_module fields. The second is to >> allow >> the conversion of function definitions from taking struct mod >> parameters to >> accepting struct boot_module as needed when a transitioned field will be >> accessed. >> >> Signed-off-by: Daniel P. Smith <dpsmith@apertussolutions.com> >> --- >> xen/arch/x86/setup.c | 61 ++++++++++++++++++++++++-------------------- >> 1 file changed, 34 insertions(+), 27 deletions(-) >> >> diff --git a/xen/arch/x86/setup.c b/xen/arch/x86/setup.c >> index dd82ca3d43e2..ba4bee6b93af 100644 >> --- a/xen/arch/x86/setup.c >> +++ b/xen/arch/x86/setup.c >> @@ -1341,15 +1341,15 @@ void asmlinkage __init noreturn >> __start_xen(unsigned long mbi_p) >> set_kexec_crash_area_size((u64)nr_pages << PAGE_SHIFT); >> kexec_reserve_area(); >> - initial_images = mod; >> + initial_images = bi->mods[0].mod; > > Isn't this wrong? > mod is the array of module_t * of *all* modules, but bi->mods[0].mod is > a single module_t *? No it is not wrong: bi->mods[0].mod == __va(mbi->mods_addr)[0] While the modules themselves get relocated, the location of the array of module_t never change. The question does give me pause to double check the patch ordering, just to be sure that mod_start and mod_end are correct up until we transition to using the boot_module fields. >> for ( i = 0; !efi_enabled(EFI_LOADER) && i < bi->nr_modules; i++ ) >> { >> - if ( mod[i].mod_start & (PAGE_SIZE - 1) ) >> + if ( bi->mods[i].mod->mod_start & (PAGE_SIZE - 1) ) >> panic("Bootloader didn't honor module alignment >> request\n"); >> - mod[i].mod_end -= mod[i].mod_start; >> - mod[i].mod_start >>= PAGE_SHIFT; >> - mod[i].reserved = 0; >> + bi->mods[i].mod->mod_end -= bi->mods[i].mod->mod_start; >> + bi->mods[i].mod->mod_start >>= PAGE_SHIFT; >> + bi->mods[i].mod->reserved = 0; >> } >> /* > >> @@ -1509,13 +1510,15 @@ void asmlinkage __init noreturn >> __start_xen(unsigned long mbi_p) >> #endif >> } >> - if ( bi->mods[0].headroom && !mod->reserved ) >> + if ( bi->mods[0].headroom && !bi->mods[0].mod->reserved ) >> panic("Not enough memory to relocate the dom0 kernel image\n"); >> for ( i = 0; i < bi->nr_modules; ++i ) >> { >> - uint64_t s = (uint64_t)mod[i].mod_start << PAGE_SHIFT; >> + uint64_t s = (uint64_t)bi->mods[i].mod->mod_start >> + << PAGE_SHIFT; > > pfn_to_paddr() ? Yep, missed one ( ._.) >> - reserve_e820_ram(&boot_e820, s, s + PAGE_ALIGN(mod[i].mod_end)); >> + reserve_e820_ram(&boot_e820, s, >> + s + PAGE_ALIGN(bi->mods[i].mod->mod_end)); >> } >> if ( !xen_phys_start ) >> @@ -1593,8 +1596,9 @@ void asmlinkage __init noreturn >> __start_xen(unsigned long mbi_p) >> map_e = boot_e820.map[j].addr + boot_e820.map[j].size; >> for ( j = 0; j < bi->nr_modules; ++j ) >> { >> - uint64_t end = pfn_to_paddr(mod[j].mod_start) + >> - mod[j].mod_end; >> + uint64_t end = pfn_to_paddr( >> + bi->mods[j].mod->mod_start) + >> + bi->mods[j].mod->mod_end; > > I think you want a different indent. I think > uint64_t end = pfn_to_paddr(bi->mods[j].mod->mod_start) > > will all fit on one line (indented all the way). (Thunderbird makes it > difficult me to send indented.) Yes, it will fit on one line without the '+', but I believe one of the many unwritten coding style rules is that the '+' stays with the LHS, so I wrapped the LHS with the '+'. >> if ( map_e < end ) >> map_e = end; >> @@ -1668,11 +1672,13 @@ void asmlinkage __init noreturn >> __start_xen(unsigned long mbi_p) >> for ( i = 0; i < bi->nr_modules; ++i ) >> { >> - set_pdx_range(mod[i].mod_start, >> - mod[i].mod_start + PFN_UP(mod[i].mod_end)); >> - map_pages_to_xen((unsigned long)mfn_to_virt(mod[i].mod_start), >> - _mfn(mod[i].mod_start), >> - PFN_UP(mod[i].mod_end), PAGE_HYPERVISOR); >> + set_pdx_range(bi->mods[i].mod->mod_start, >> + bi->mods[i].mod->mod_start + >> + PFN_UP(bi->mods[i].mod->mod_end)); >> + map_pages_to_xen( >> + (unsigned long)mfn_to_virt(bi->mods[i].mod->mod_start), > > map_pages_to_xen((unsigned long)maddr_to_virt(bi->mods[i].start), > > All fits on one line. If it does, I will bring it back up. v/r, dps
On 09.10.2024 16:23, Daniel P. Smith wrote: > On 10/7/24 15:34, Jason Andryuk wrote: >> On 2024-10-06 17:49, Daniel P. Smith wrote: >>> --- a/xen/arch/x86/setup.c >>> +++ b/xen/arch/x86/setup.c >>> @@ -1341,15 +1341,15 @@ void asmlinkage __init noreturn >>> __start_xen(unsigned long mbi_p) >>> set_kexec_crash_area_size((u64)nr_pages << PAGE_SHIFT); >>> kexec_reserve_area(); >>> - initial_images = mod; >>> + initial_images = bi->mods[0].mod; >> >> Isn't this wrong? >> mod is the array of module_t * of *all* modules, but bi->mods[0].mod is >> a single module_t *? > > No it is not wrong: > bi->mods[0].mod == __va(mbi->mods_addr)[0] Yet as it's seemingly wrong, a comment appears to be necessary. Jan
On 10/9/24 10:29, Jan Beulich wrote: > On 09.10.2024 16:23, Daniel P. Smith wrote: >> On 10/7/24 15:34, Jason Andryuk wrote: >>> On 2024-10-06 17:49, Daniel P. Smith wrote: >>>> --- a/xen/arch/x86/setup.c >>>> +++ b/xen/arch/x86/setup.c >>>> @@ -1341,15 +1341,15 @@ void asmlinkage __init noreturn >>>> __start_xen(unsigned long mbi_p) >>>> set_kexec_crash_area_size((u64)nr_pages << PAGE_SHIFT); >>>> kexec_reserve_area(); >>>> - initial_images = mod; >>>> + initial_images = bi->mods[0].mod; >>> >>> Isn't this wrong? >>> mod is the array of module_t * of *all* modules, but bi->mods[0].mod is >>> a single module_t *? >> >> No it is not wrong: >> bi->mods[0].mod == __va(mbi->mods_addr)[0] > > Yet as it's seemingly wrong, a comment appears to be necessary. That is fair, will add comment. v/r, dps
On 06.10.2024 23:49, Daniel P. Smith wrote: > @@ -2061,8 +2067,9 @@ void asmlinkage __init noreturn __start_xen(unsigned long mbi_p) > * We're going to setup domain0 using the module(s) that we stashed safely > * above our heap. The second module, if present, is an initrd ramdisk. > */ > - dom0 = create_dom0(mod, bi->mods[0].headroom, > - initrdidx < bi->nr_modules ? mod + initrdidx : NULL, > + dom0 = create_dom0(bi->mods[0].mod, bi->mods[0].headroom, > + initrdidx < bi->nr_modules ? > + bi->mods[initrdidx].mod : NULL, See an earlier comment regarding wrapped ?:. We certainly never have indentation levels of 5 blanks. Jan
On 10/9/24 11:29, Jan Beulich wrote: > On 06.10.2024 23:49, Daniel P. Smith wrote: >> @@ -2061,8 +2067,9 @@ void asmlinkage __init noreturn __start_xen(unsigned long mbi_p) >> * We're going to setup domain0 using the module(s) that we stashed safely >> * above our heap. The second module, if present, is an initrd ramdisk. >> */ >> - dom0 = create_dom0(mod, bi->mods[0].headroom, >> - initrdidx < bi->nr_modules ? mod + initrdidx : NULL, >> + dom0 = create_dom0(bi->mods[0].mod, bi->mods[0].headroom, >> + initrdidx < bi->nr_modules ? >> + bi->mods[initrdidx].mod : NULL, > > See an earlier comment regarding wrapped ?:. We certainly never have > indentation levels of 5 blanks. Can format similar as earlier, and just fyi, that is a stray space from wrapping and not a five space indent. v/r, dps
diff --git a/xen/arch/x86/setup.c b/xen/arch/x86/setup.c index dd82ca3d43e2..ba4bee6b93af 100644 --- a/xen/arch/x86/setup.c +++ b/xen/arch/x86/setup.c @@ -1341,15 +1341,15 @@ void asmlinkage __init noreturn __start_xen(unsigned long mbi_p) set_kexec_crash_area_size((u64)nr_pages << PAGE_SHIFT); kexec_reserve_area(); - initial_images = mod; + initial_images = bi->mods[0].mod; for ( i = 0; !efi_enabled(EFI_LOADER) && i < bi->nr_modules; i++ ) { - if ( mod[i].mod_start & (PAGE_SIZE - 1) ) + if ( bi->mods[i].mod->mod_start & (PAGE_SIZE - 1) ) panic("Bootloader didn't honor module alignment request\n"); - mod[i].mod_end -= mod[i].mod_start; - mod[i].mod_start >>= PAGE_SHIFT; - mod[i].reserved = 0; + bi->mods[i].mod->mod_end -= bi->mods[i].mod->mod_start; + bi->mods[i].mod->mod_start >>= PAGE_SHIFT; + bi->mods[i].mod->reserved = 0; } /* @@ -1360,6 +1360,7 @@ void asmlinkage __init noreturn __start_xen(unsigned long mbi_p) if ( xen_phys_start ) { + unsigned int xen = bi->nr_modules; relocated = true; /* @@ -1367,8 +1368,8 @@ void asmlinkage __init noreturn __start_xen(unsigned long mbi_p) * respective reserve_e820_ram() invocation below. No need to * query efi_boot_mem_unused() here, though. */ - mod[bi->nr_modules].mod_start = virt_to_mfn(_stext); - mod[bi->nr_modules].mod_end = __2M_rwdata_end - _stext; + bi->mods[xen].mod->mod_start = virt_to_mfn(_stext); + bi->mods[xen].mod->mod_end = __2M_rwdata_end - _stext; } bi->mods[0].headroom = @@ -1462,9 +1463,9 @@ void asmlinkage __init noreturn __start_xen(unsigned long mbi_p) struct boot_module *bm = &bi->mods[j]; unsigned long size; - size = PAGE_ALIGN(bm->headroom + mod[j].mod_end); + size = PAGE_ALIGN(bm->headroom + bm->mod->mod_end); - if ( mod[j].reserved ) + if ( bi->mods[j].mod->reserved ) continue; /* Don't overlap with other modules (or Xen itself). */ @@ -1476,14 +1477,14 @@ void asmlinkage __init noreturn __start_xen(unsigned long mbi_p) if ( s < end && (bm->headroom || - ((end - size) >> PAGE_SHIFT) > mod[j].mod_start) ) + ((end - size) >> PAGE_SHIFT) > bm->mod->mod_start) ) { move_memory(end - size + bm->headroom, - (uint64_t)mod[j].mod_start << PAGE_SHIFT, - mod[j].mod_end); - mod[j].mod_start = (end - size) >> PAGE_SHIFT; - mod[j].mod_end += bm->headroom; - mod[j].reserved = 1; + (uint64_t)bm->mod->mod_start << PAGE_SHIFT, + bm->mod->mod_end); + bm->mod->mod_start = (end - size) >> PAGE_SHIFT; + bm->mod->mod_end += bm->headroom; + bm->mod->reserved = 1; } } @@ -1509,13 +1510,15 @@ void asmlinkage __init noreturn __start_xen(unsigned long mbi_p) #endif } - if ( bi->mods[0].headroom && !mod->reserved ) + if ( bi->mods[0].headroom && !bi->mods[0].mod->reserved ) panic("Not enough memory to relocate the dom0 kernel image\n"); for ( i = 0; i < bi->nr_modules; ++i ) { - uint64_t s = (uint64_t)mod[i].mod_start << PAGE_SHIFT; + uint64_t s = (uint64_t)bi->mods[i].mod->mod_start + << PAGE_SHIFT; - reserve_e820_ram(&boot_e820, s, s + PAGE_ALIGN(mod[i].mod_end)); + reserve_e820_ram(&boot_e820, s, + s + PAGE_ALIGN(bi->mods[i].mod->mod_end)); } if ( !xen_phys_start ) @@ -1593,8 +1596,9 @@ void asmlinkage __init noreturn __start_xen(unsigned long mbi_p) map_e = boot_e820.map[j].addr + boot_e820.map[j].size; for ( j = 0; j < bi->nr_modules; ++j ) { - uint64_t end = pfn_to_paddr(mod[j].mod_start) + - mod[j].mod_end; + uint64_t end = pfn_to_paddr( + bi->mods[j].mod->mod_start) + + bi->mods[j].mod->mod_end; if ( map_e < end ) map_e = end; @@ -1668,11 +1672,13 @@ void asmlinkage __init noreturn __start_xen(unsigned long mbi_p) for ( i = 0; i < bi->nr_modules; ++i ) { - set_pdx_range(mod[i].mod_start, - mod[i].mod_start + PFN_UP(mod[i].mod_end)); - map_pages_to_xen((unsigned long)mfn_to_virt(mod[i].mod_start), - _mfn(mod[i].mod_start), - PFN_UP(mod[i].mod_end), PAGE_HYPERVISOR); + set_pdx_range(bi->mods[i].mod->mod_start, + bi->mods[i].mod->mod_start + + PFN_UP(bi->mods[i].mod->mod_end)); + map_pages_to_xen( + (unsigned long)mfn_to_virt(bi->mods[i].mod->mod_start), + _mfn(bi->mods[i].mod->mod_start), + PFN_UP(bi->mods[i].mod->mod_end), PAGE_HYPERVISOR); } #ifdef CONFIG_KEXEC @@ -2061,8 +2067,9 @@ void asmlinkage __init noreturn __start_xen(unsigned long mbi_p) * We're going to setup domain0 using the module(s) that we stashed safely * above our heap. The second module, if present, is an initrd ramdisk. */ - dom0 = create_dom0(mod, bi->mods[0].headroom, - initrdidx < bi->nr_modules ? mod + initrdidx : NULL, + dom0 = create_dom0(bi->mods[0].mod, bi->mods[0].headroom, + initrdidx < bi->nr_modules ? + bi->mods[initrdidx].mod : NULL, kextra, bi->loader); if ( !dom0 ) panic("Could not set up DOM0 guest OS\n");
To allow a slow conversion of x86 over to struct boot_module, start with replacing all references to struct mod to the early_mod element of struct boot_module. These serves twofold, first to allow the incremental transition from struct mod fields to struct boot_module fields. The second is to allow the conversion of function definitions from taking struct mod parameters to accepting struct boot_module as needed when a transitioned field will be accessed. Signed-off-by: Daniel P. Smith <dpsmith@apertussolutions.com> --- xen/arch/x86/setup.c | 61 ++++++++++++++++++++++++-------------------- 1 file changed, 34 insertions(+), 27 deletions(-)