Message ID | 1372298434-20220-2-git-send-email-haokexin@gmail.com (mailing list archive) |
---|---|
State | Superseded |
Headers | show |
On Thu, Jun 27, 2013 at 02:58:34PM -0500, Scott Wood wrote: > On 06/26/2013 09:00:33 PM, Kevin Hao wrote: > >This is based on the codes in the head_44x.S. Since we always align to > >256M before mapping the PAGE_OFFSET for a relocatable kernel, we also > >change the init tlb map to 256M size. > > Why 256M? For two reasons: 1. This is the size which both e500v1 and e500v2 support. 2. Since we always use the PAGE_OFFSET as 0xc0000000, the 256M is max alignment value we can use for this virtual address. > > This tightens the alignment requirement for dynamic memstart. Yes. But since RELOCATABLE is a superset of DYNAMIC_MEMSTART, we can always use RELOCATABLE instead of DYNAMIC_MEMSTART for fsl booke board in any cases. So DYNAMIC_MEMSTART will seem not so useful after we enable this feature. > And > what about boards with less than 256 MiB of RAM? It should be fine. We just create the map in the tlb. The MM still use the real size of memory. > > >@@ -176,6 +176,8 @@ skpinv: addi r6,r6,1 /* Increment */ > > /* 7. Jump to KERNELBASE mapping */ > > lis r6,(KERNELBASE & ~0xfff)@h > > ori r6,r6,(KERNELBASE & ~0xfff)@l > >+ rlwinm r7,r25,0,4,31 > >+ add r6,r7,r6 > > Please consider using the more readable form of rlwinm/rlwimi: > > rlwinm r7,r25,0,0x0fffffff Sure. > > > #elif defined(ENTRY_MAPPING_KEXEC_SETUP) > > /* > >diff --git a/arch/powerpc/kernel/head_fsl_booke.S > >b/arch/powerpc/kernel/head_fsl_booke.S > >index d10a7ca..c3b4c8e53 100644 > >--- a/arch/powerpc/kernel/head_fsl_booke.S > >+++ b/arch/powerpc/kernel/head_fsl_booke.S > >@@ -83,10 +83,43 @@ _ENTRY(_start); > > andc r31,r20,r18 /* r31 = page base */ > > or r31,r31,r19 /* r31 = devtree phys addr */ > > mfspr r30,SPRN_MAS7 > >- > >- li r25,0 /* phys kernel start (low) */ > > li r24,0 /* CPU number */ > >- li r23,0 /* phys kernel start (high) */ > >+ > >+#ifdef CONFIG_RELOCATABLE > >+ bl 0f /* Get our runtime address */ > >+0: mflr r3 /* Make it accessible */ > >+ addis r3,r3,(_stext - 0b)@ha > >+ addi r3,r3,(_stext - 0b)@l /* Get our current runtime base */ > >+ > >+ /* Translate _stext address to physical, save in r23/r25 */ > >+ tlbsx 0,r3 /* must succeed */ > >+ > >+ mfspr r16,SPRN_MAS1 > >+ mfspr r20,SPRN_MAS3 > >+ rlwinm r17,r16,25,0x1f /* r17 = log2(page size) */ > >+ li r18,1024 > >+ slw r18,r18,r17 /* r18 = page size */ > >+ addi r18,r18,-1 > >+ and r19,r3,r18 /* r19 = page offset */ > >+ andc r25,r20,r18 /* r25 = page base */ > >+ or r25,r25,r19 /* r25 = _stext phys addr */ > >+ mfspr r23,SPRN_MAS7 > > This duplicates the code for finding the device tree physical > address... maybe factor it out into a function? Sure. > > >@@ -197,7 +230,58 @@ _ENTRY(__early_start) > > > > bl early_init > > > >-#ifdef CONFIG_DYNAMIC_MEMSTART > >+#ifdef CONFIG_RELOCATABLE > >+ /* > >+ * Relocatable kernel support based on processing of dynamic > >+ * relocation entries. > >+ * > >+ * r25/r23 will contain RPN/ERPN for the start address of memory > > The start of memory or the start of the kernel? Should be the start of the kernel. Will fix the comment. > > >+ */ > >+ lis r3,kernstart_addr@ha > >+ la r3,kernstart_addr@l(r3) > >+ > >+#ifdef CONFIG_PHYS_64BIT > >+ stw r23,0(r3) > >+ stw r25,4(r3) > >+#else > >+ stw r25,0(r3) > >+#endif > > This part looks the same for relocatable and dynamic memstart -- can > you avoid duplicating? OK. > > >+ /* > >+ * Compute the virt_phys_offset : > >+ * virt_phys_offset = stext.run - kernstart_addr > >+ * > >+ * stext.run = (KERNELBASE & ~0xfffffff) + (kernstart_addr & > >0xfffffff) > >+ * When we relocate, we have : > >+ * > >+ * (kernstart_addr & 0xfffffff) = (stext.run & 0xfffffff) > >+ * > >+ * hence: > >+ * virt_phys_offset = (KERNELBASE & ~0xfffffff) - > >+ * (kernstart_addr & ~0xfffffff) > >+ * > >+ */ > >+ > >+ /* KERNELBASE&~0xfffffff => (r4,r5) */ > > >+ li r4, 0 /* higer 32bit */ > >+ lis r5,KERNELBASE@h > > Please be consistent with whitespace. Fixed. Thanks, Kevin > > -Scott
On 06/27/2013 08:36:37 PM, Kevin Hao wrote: > On Thu, Jun 27, 2013 at 02:58:34PM -0500, Scott Wood wrote: > > On 06/26/2013 09:00:33 PM, Kevin Hao wrote: > > >This is based on the codes in the head_44x.S. Since we always > align to > > >256M before mapping the PAGE_OFFSET for a relocatable kernel, we > also > > >change the init tlb map to 256M size. > > > > Why 256M? > > For two reasons: > 1. This is the size which both e500v1 and e500v2 support. > 2. Since we always use the PAGE_OFFSET as 0xc0000000, the 256M is > max alignment value we can use for this virtual address. 0xc0000000 is 1G-aligned, so I don't see why 256M is the maximum (after verifying that enough memory is present with the right alignment, of course). The TLB1 savings would probably not be enough to justify figuring that out, though. -Scott
On Thu, Jun 27, 2013 at 08:47:27PM -0500, Scott Wood wrote: > On 06/27/2013 08:36:37 PM, Kevin Hao wrote: > >On Thu, Jun 27, 2013 at 02:58:34PM -0500, Scott Wood wrote: > >> On 06/26/2013 09:00:33 PM, Kevin Hao wrote: > >> >This is based on the codes in the head_44x.S. Since we always > >align to > >> >256M before mapping the PAGE_OFFSET for a relocatable kernel, > >we also > >> >change the init tlb map to 256M size. > >> > >> Why 256M? > > > >For two reasons: > > 1. This is the size which both e500v1 and e500v2 support. > > 2. Since we always use the PAGE_OFFSET as 0xc0000000, the 256M is > > max alignment value we can use for this virtual address. > > Is there any reason why 64M won't continue to work here? Yes. In general we would map the 0 ~ 256M memory region in the first tlb1 entry. If we align to 64M, the relocatable kernel would not work if loaded above 64M memory. For example, if we load a relocatable kernel at 64M memory, we will relocate it as: __pa(PAGE_OFFSET) = 0x4000000 But in map_mem_in_cams function, it will create a memory map as: __pa(PAGE_OFFSET) = 0x0 The kernel will definitely not work in this case. > > >> This tightens the alignment requirement for dynamic memstart. > > > >Yes. But since RELOCATABLE is a superset of DYNAMIC_MEMSTART, we > >can always > >use RELOCATABLE instead of DYNAMIC_MEMSTART for fsl booke board in > >any cases. > > The extra flexibility of RELOCATABLE may help some use cases, but > you'd still require the entire 256M naturally aligned region > containing the kernel to be present and owned by this instance of > Linux. > > >So DYNAMIC_MEMSTART will seem not so useful after we enable this > >feature. > > Then why doesn't this patch remove it? According to the Kconfig it is still used by 44x. And maybe someone still want to use this relocation method. > > >> And > >> what about boards with less than 256 MiB of RAM? > > > >It should be fine. We just create the map in the tlb. The MM still use > >the real size of memory. > > No, you must not map anything that is not present with a mapping > that is executable and/or not guarded, or you could get speculative > accesses to who-knows-what. Yes, there may be speculative access in this case. > Even if RAM is present there but owned > by some other entity, you could be creating illegal aliases if that > other entity mapped it cache-inhibited or similar. Fair enough. So it seems error prone if we map this 256M memory region blindly. But if we don't do this, it seems that we have to do twice relocation. The first time we just align to a predefined value (64M for example), and then parse the device tree and get the real memstart_addr. After that we should relocate the kernel to the real start address. It seems a little complicated. Do you have any better ideas? Thanks, Kevin > > -Scott
On Thu, Jun 27, 2013 at 08:52:20PM -0500, Scott Wood wrote: > On 06/27/2013 08:36:37 PM, Kevin Hao wrote: > >On Thu, Jun 27, 2013 at 02:58:34PM -0500, Scott Wood wrote: > >> On 06/26/2013 09:00:33 PM, Kevin Hao wrote: > >> >This is based on the codes in the head_44x.S. Since we always > >align to > >> >256M before mapping the PAGE_OFFSET for a relocatable kernel, > >we also > >> >change the init tlb map to 256M size. > >> > >> Why 256M? > > > >For two reasons: > > 1. This is the size which both e500v1 and e500v2 support. > > 2. Since we always use the PAGE_OFFSET as 0xc0000000, the 256M is > > max alignment value we can use for this virtual address. > > 0xc0000000 is 1G-aligned, so I don't see why 256M is the maximum > (after verifying that enough memory is present with the right > alignment, of course). The TLB1 savings would probably not be > enough to justify figuring that out, though. Sorry, I didn't make myself clear. Yes, the 0xc0000000 is 1G-aligned. For a 32bit kernel we only have a 1G memory region for the kernel space, but we can't use all of it for direct map. So we always set the __max_low_memory to 0x30000000. And for e500 core, it doesn't support 512M page size. So the 256M is the max page size we can use for a 32bit kernel. And since we always cover the boot code in the first tlb entry, we then use the 256M as the alignment value for the relocatable kernel. Thanks, Kevin > > -Scott
On Mon, Jul 01, 2013 at 07:30:45PM -0500, Scott Wood wrote: > On 06/30/2013 02:33:10 AM, Kevin Hao wrote: > >On Thu, Jun 27, 2013 at 08:47:27PM -0500, Scott Wood wrote: > >> On 06/27/2013 08:36:37 PM, Kevin Hao wrote: > >> >On Thu, Jun 27, 2013 at 02:58:34PM -0500, Scott Wood wrote: > >> >> On 06/26/2013 09:00:33 PM, Kevin Hao wrote: > >> >> >This is based on the codes in the head_44x.S. Since we always > >> >align to > >> >> >256M before mapping the PAGE_OFFSET for a relocatable kernel, > >> >we also > >> >> >change the init tlb map to 256M size. > >> >> > >> >> Why 256M? > >> > > >> >For two reasons: > >> > 1. This is the size which both e500v1 and e500v2 support. > >> > 2. Since we always use the PAGE_OFFSET as 0xc0000000, the 256M is > >> > max alignment value we can use for this virtual address. > >> > >> Is there any reason why 64M won't continue to work here? > > > >Yes. In general we would map the 0 ~ 256M memory region in the first > >tlb1 entry. If we align to 64M, the relocatable kernel would not work > >if loaded above 64M memory. For example, if we load a relocatable > >kernel > >at 64M memory, we will relocate it as: > > __pa(PAGE_OFFSET) = 0x4000000 > > > >But in map_mem_in_cams function, it will create a memory map as: > > __pa(PAGE_OFFSET) = 0x0 > > > >The kernel will definitely not work in this case. > > That's a problem with map_mem_in_cams(), as discussed in the thread > on other patch. Perhaps fully solving those problems is not > worthwhile at this time, but we should at least be able to determine > the TLB size automatically based on the alignment of the address > you're trying to map. 64M would be used unless (address & (256M - > 1)) >= 64M. I hope we can continue to assume the kernel won't cross > a 64M boundary. No. The problem is we don't know the physical address of the start of lowmem at booting. So we have to align to physical address (phys1) blindly and map the PAGE_OFFSET from there. Then once we get the physical address (phys2) of the start of lowmem from the device tree later, we will map the PAGE_OFFSET to the start of lowmem. If the phys1 is not equal to phys2, we get a problem. The reasons that we can't still map the PAGE_OFFSET to phys1 in the map_mem_in_cams() are: * if phys1 > phys2, then this will waste the memory between phys2 ~ phys1. * if phys1 < phys2, then we map a memory region which don't belong to this kernel. > > >> >> This tightens the alignment requirement for dynamic memstart. > >> > > >> >Yes. But since RELOCATABLE is a superset of DYNAMIC_MEMSTART, we > >> >can always > >> >use RELOCATABLE instead of DYNAMIC_MEMSTART for fsl booke board in > >> >any cases. > >> > >> The extra flexibility of RELOCATABLE may help some use cases, but > >> you'd still require the entire 256M naturally aligned region > >> containing the kernel to be present and owned by this instance of > >> Linux. > >> > >> >So DYNAMIC_MEMSTART will seem not so useful after we enable this > >> >feature. > >> > >> Then why doesn't this patch remove it? > > > >According to the Kconfig it is still used by 44x. > > RELOCATABLE appears to be supported on 44x, and is what CRASH_DUMP > uses on 44x. The kdump kernel on 44x also use the RELOCATABLE method. > > >And maybe someone still want to use this relocation method. > > Then you don't get to dismiss claims that you're changing > DYNAMIC_MEMSTART alignment requirements by saying that RELOCATABLE > is a strict superset. :-) Given the requirement that the kernel be > in the first TLB entry, though, using RELOCATABLE rather than > DYNAMIC_MEMSTART doesn't fix the alignment problem. > > I don't think it makes sense to keep both mechanisms around unless > there's some obvious reason to prefer DYNAMIC_MEMSTART. The DYNAMIC_MEMSTART still can be used for such as AMP kernel. It does have a more small footprint than RELOCATABLE and also doesn't have the overhead of the relocation. So I don't want to drop it in a rush. Thanks, Kevin > > >> >> And > >> >> what about boards with less than 256 MiB of RAM? > >> > > >> >It should be fine. We just create the map in the tlb. The MM > >still use > >> >the real size of memory. > >> > >> No, you must not map anything that is not present with a mapping > >> that is executable and/or not guarded, or you could get speculative > >> accesses to who-knows-what. > > > >Yes, there may be speculative access in this case. > > > >> Even if RAM is present there but owned > >> by some other entity, you could be creating illegal aliases if that > >> other entity mapped it cache-inhibited or similar. > > > >Fair enough. So it seems error prone if we map this 256M memory region > >blindly. But if we don't do this, it seems that we have to do > >twice relocation. > >The first time we just align to a predefined value (64M for > >example), and > >then parse the device tree and get the real memstart_addr. After > >that we > >should relocate the kernel to the real start address. It seems a > >little > >complicated. Do you have any better ideas? > > This seems like the proper way to address it, assuming we're > unwilling to map the kernel image somewhere other than the normal > lowmem mapping (and I think we're unwilling, given how tight the > address space is on 32-bit, and the intrusiveness of the change). > The dynamic determination of 64M versus 256M could be an acceptable > alternative though, if we're OK with not supporting arbitrary > relocatable scenarios, but just those that are either needed by > kdump, or supported by current kernels (with DYNAMIC_MEMSTART, or > just starting at zero with less than 256M of RAM). If we go that > route, the limitations should be documented. > > -Scott
On Tue, Jul 02, 2013 at 05:39:18PM -0500, Scott Wood wrote: > On 07/01/2013 10:24:47 PM, Kevin Hao wrote: > >On Mon, Jul 01, 2013 at 07:30:45PM -0500, Scott Wood wrote: > >> On 06/30/2013 02:33:10 AM, Kevin Hao wrote: > >> >On Thu, Jun 27, 2013 at 08:47:27PM -0500, Scott Wood wrote: > >> >> On 06/27/2013 08:36:37 PM, Kevin Hao wrote: > >> >> >On Thu, Jun 27, 2013 at 02:58:34PM -0500, Scott Wood wrote: > >> >> >> On 06/26/2013 09:00:33 PM, Kevin Hao wrote: > >> >> >> >This is based on the codes in the head_44x.S. Since we always > >> >> >align to > >> >> >> >256M before mapping the PAGE_OFFSET for a relocatable kernel, > >> >> >we also > >> >> >> >change the init tlb map to 256M size. > >> >> >> > >> >> >> Why 256M? > >> >> > > >> >> >For two reasons: > >> >> > 1. This is the size which both e500v1 and e500v2 support. > >> >> > 2. Since we always use the PAGE_OFFSET as 0xc0000000, the > >256M is > >> >> > max alignment value we can use for this virtual address. > >> >> > >> >> Is there any reason why 64M won't continue to work here? > >> > > >> >Yes. In general we would map the 0 ~ 256M memory region in the > >first > >> >tlb1 entry. If we align to 64M, the relocatable kernel would > >not work > >> >if loaded above 64M memory. For example, if we load a relocatable > >> >kernel > >> >at 64M memory, we will relocate it as: > >> > __pa(PAGE_OFFSET) = 0x4000000 > >> > > >> >But in map_mem_in_cams function, it will create a memory map as: > >> > __pa(PAGE_OFFSET) = 0x0 > >> > > >> >The kernel will definitely not work in this case. > >> > >> That's a problem with map_mem_in_cams(), as discussed in the thread > >> on other patch. Perhaps fully solving those problems is not > >> worthwhile at this time, but we should at least be able to determine > >> the TLB size automatically based on the alignment of the address > >> you're trying to map. 64M would be used unless (address & (256M - > >> 1)) >= 64M. I hope we can continue to assume the kernel won't cross > >> a 64M boundary. > > > >No. The problem is we don't know the physical address of the start of > >lowmem at booting. So we have to align to physical address (phys1) > >blindly > >and map the PAGE_OFFSET from there. Then once we get the physical > >address > >(phys2) of the start of lowmem from the device tree later, we will > >map the > >PAGE_OFFSET to the start of lowmem. If the phys1 is not equal to > >phys2, > >we get a problem. > > How would you get phys1 != phys2, unless the kernel begins in a > 256M-aligned region other than the first (which you said is already > not supported)? Yes, this is the only case which phys1 != phys2 if we align to 256M. I plan to also fix this in the next version. > > If (phys1 & (256M - 1)) < 64M, then you'd get the same phys2 > regardless of whether you align it to 64M or 256M. > Otherwise, we use a 256M page which is what you're already doing. Yes, you are right. I am just trying to say we will run into problem when loading a kernel between 64M ~ 256M if we don't align to 256M. > > >> >And maybe someone still want to use this relocation method. > >> > >> Then you don't get to dismiss claims that you're changing > >> DYNAMIC_MEMSTART alignment requirements by saying that RELOCATABLE > >> is a strict superset. :-) Given the requirement that the kernel be > >> in the first TLB entry, though, using RELOCATABLE rather than > >> DYNAMIC_MEMSTART doesn't fix the alignment problem. > >> > >> I don't think it makes sense to keep both mechanisms around unless > >> there's some obvious reason to prefer DYNAMIC_MEMSTART. > > > >The DYNAMIC_MEMSTART still can be used for such as AMP kernel. It > >does have > >a more small footprint than RELOCATABLE and also doesn't have the > >overhead > >of the relocation. So I don't want to drop it in a rush. > > How much overhead (space and time) is this really? The following is the additional sections when relocatable is enabled for a p2020rdb board. section size .dynsym 000007f0 .dynstr 00000926 .dynamic 00000080 .hash 00000388 .interp 00000011 .rela.dyn 00215250 The time for the relocation is about 32ms on a p2020rdb board. > > It will keep the code (and especially the diff) simpler to have this > replace DYNAMIC_MEMSTART rather than add to it. OK. If you think that the above overhead is acceptable, I can drop the DYNAMIC_MEMSTART in the next version. Thanks, Kevin > > -Scott
On 07/02/2013 10:00:44 PM, Kevin Hao wrote: > On Tue, Jul 02, 2013 at 05:39:18PM -0500, Scott Wood wrote: > > How much overhead (space and time) is this really? > > The following is the additional sections when relocatable is enabled > for > a p2020rdb board. > section size > .dynsym 000007f0 > .dynstr 00000926 > .dynamic 00000080 > .hash 00000388 > .interp 00000011 > .rela.dyn 00215250 > > The time for the relocation is about 32ms on a p2020rdb board. Hmm... more relocations than I expected. What percentage is this of the total image size? -Scott
On Wed, Jul 03, 2013 at 03:38:27PM -0500, Scott Wood wrote: > On 07/02/2013 10:00:44 PM, Kevin Hao wrote: > >On Tue, Jul 02, 2013 at 05:39:18PM -0500, Scott Wood wrote: > >> How much overhead (space and time) is this really? > > > >The following is the additional sections when relocatable is > >enabled for > >a p2020rdb board. > > section size > > .dynsym 000007f0 > > .dynstr 00000926 > > .dynamic 00000080 > > .hash 00000388 > > .interp 00000011 > > .rela.dyn 00215250 > > > >The time for the relocation is about 32ms on a p2020rdb board. > > Hmm... more relocations than I expected. What percentage is this of > the total image size? The size of vmlinux.bin is about 10M. The percentage of the relocation section is about 20%. But look on the bright side of thing, all the relocation stuff are in init section and should be discarded at runtime. :-) Thanks, Kevin > > -Scott
On 07/03/2013 08:08:18 PM, Kevin Hao wrote: > On Wed, Jul 03, 2013 at 03:38:27PM -0500, Scott Wood wrote: > > On 07/02/2013 10:00:44 PM, Kevin Hao wrote: > > >On Tue, Jul 02, 2013 at 05:39:18PM -0500, Scott Wood wrote: > > >> How much overhead (space and time) is this really? > > > > > >The following is the additional sections when relocatable is > > >enabled for > > >a p2020rdb board. > > > section size > > > .dynsym 000007f0 > > > .dynstr 00000926 > > > .dynamic 00000080 > > > .hash 00000388 > > > .interp 00000011 > > > .rela.dyn 00215250 > > > > > >The time for the relocation is about 32ms on a p2020rdb board. > > > > Hmm... more relocations than I expected. What percentage is this of > > the total image size? > > The size of vmlinux.bin is about 10M. The percentage of the relocation > section is about 20%. But look on the bright side of thing, all the > relocation > stuff are in init section and should be discarded at runtime. :-) That doesn't reduce the space it takes up in flash... -Scott
On Mon, Jul 08, 2013 at 11:48:25AM -0500, Scott Wood wrote: > On 07/03/2013 08:08:18 PM, Kevin Hao wrote: > >On Wed, Jul 03, 2013 at 03:38:27PM -0500, Scott Wood wrote: > >> On 07/02/2013 10:00:44 PM, Kevin Hao wrote: > >> >On Tue, Jul 02, 2013 at 05:39:18PM -0500, Scott Wood wrote: > >> >> How much overhead (space and time) is this really? > >> > > >> >The following is the additional sections when relocatable is > >> >enabled for > >> >a p2020rdb board. > >> > section size > >> > .dynsym 000007f0 > >> > .dynstr 00000926 > >> > .dynamic 00000080 > >> > .hash 00000388 > >> > .interp 00000011 > >> > .rela.dyn 00215250 > >> > > >> >The time for the relocation is about 32ms on a p2020rdb board. > >> > >> Hmm... more relocations than I expected. What percentage is this of > >> the total image size? > > > >The size of vmlinux.bin is about 10M. The percentage of the relocation > >section is about 20%. But look on the bright side of thing, all > >the relocation > >stuff are in init section and should be discarded at runtime. :-) > > That doesn't reduce the space it takes up in flash... Yes. But since we always use a compressed uImage kernel in flash, the increasing size of the image would shrink to about 500K. Thanks, Kevin > > -Scott
diff --git a/arch/powerpc/Kconfig b/arch/powerpc/Kconfig index c33e3ad..9eb97ac 100644 --- a/arch/powerpc/Kconfig +++ b/arch/powerpc/Kconfig @@ -866,7 +866,7 @@ config DYNAMIC_MEMSTART config RELOCATABLE bool "Build a relocatable kernel" - depends on ADVANCED_OPTIONS && FLATMEM && 44x + depends on ADVANCED_OPTIONS && FLATMEM && (44x || FSL_BOOKE) select NONSTATIC_KERNEL help This builds a kernel image that is capable of running at the diff --git a/arch/powerpc/kernel/fsl_booke_entry_mapping.S b/arch/powerpc/kernel/fsl_booke_entry_mapping.S index a92c79b..32a4b38 100644 --- a/arch/powerpc/kernel/fsl_booke_entry_mapping.S +++ b/arch/powerpc/kernel/fsl_booke_entry_mapping.S @@ -165,10 +165,10 @@ skpinv: addi r6,r6,1 /* Increment */ lis r6,0x1000 /* Set MAS0(TLBSEL) = TLB1(1), ESEL = 0 */ mtspr SPRN_MAS0,r6 lis r6,(MAS1_VALID|MAS1_IPROT)@h - ori r6,r6,(MAS1_TSIZE(BOOK3E_PAGESZ_64M))@l + ori r6,r6,(MAS1_TSIZE(BOOK3E_PAGESZ_256M))@l mtspr SPRN_MAS1,r6 - lis r6,MAS2_VAL(PAGE_OFFSET, BOOK3E_PAGESZ_64M, M_IF_SMP)@h - ori r6,r6,MAS2_VAL(PAGE_OFFSET, BOOK3E_PAGESZ_64M, M_IF_SMP)@l + lis r6,MAS2_VAL(PAGE_OFFSET, BOOK3E_PAGESZ_256M, M_IF_SMP)@h + ori r6,r6,MAS2_VAL(PAGE_OFFSET, BOOK3E_PAGESZ_256M, M_IF_SMP)@l mtspr SPRN_MAS2,r6 mtspr SPRN_MAS3,r8 tlbwe @@ -176,6 +176,8 @@ skpinv: addi r6,r6,1 /* Increment */ /* 7. Jump to KERNELBASE mapping */ lis r6,(KERNELBASE & ~0xfff)@h ori r6,r6,(KERNELBASE & ~0xfff)@l + rlwinm r7,r25,0,4,31 + add r6,r7,r6 #elif defined(ENTRY_MAPPING_KEXEC_SETUP) /* diff --git a/arch/powerpc/kernel/head_fsl_booke.S b/arch/powerpc/kernel/head_fsl_booke.S index d10a7ca..c3b4c8e53 100644 --- a/arch/powerpc/kernel/head_fsl_booke.S +++ b/arch/powerpc/kernel/head_fsl_booke.S @@ -83,10 +83,43 @@ _ENTRY(_start); andc r31,r20,r18 /* r31 = page base */ or r31,r31,r19 /* r31 = devtree phys addr */ mfspr r30,SPRN_MAS7 - - li r25,0 /* phys kernel start (low) */ li r24,0 /* CPU number */ - li r23,0 /* phys kernel start (high) */ + +#ifdef CONFIG_RELOCATABLE + bl 0f /* Get our runtime address */ +0: mflr r3 /* Make it accessible */ + addis r3,r3,(_stext - 0b)@ha + addi r3,r3,(_stext - 0b)@l /* Get our current runtime base */ + + /* Translate _stext address to physical, save in r23/r25 */ + tlbsx 0,r3 /* must succeed */ + + mfspr r16,SPRN_MAS1 + mfspr r20,SPRN_MAS3 + rlwinm r17,r16,25,0x1f /* r17 = log2(page size) */ + li r18,1024 + slw r18,r18,r17 /* r18 = page size */ + addi r18,r18,-1 + and r19,r3,r18 /* r19 = page offset */ + andc r25,r20,r18 /* r25 = page base */ + or r25,r25,r19 /* r25 = _stext phys addr */ + mfspr r23,SPRN_MAS7 + + /* + * We have the runtime (virutal) address of our base. + * We calculate our shift of offset from a 256M page. + * We could map the 256M page we belong to at PAGE_OFFSET and + * get going from there. + */ + lis r4,KERNELBASE@h + ori r4,r4,KERNELBASE@l + rlwinm r6,r25,0,4,31 /* r6 = PHYS_START % 256M */ + rlwinm r5,r4,0,4,31 /* r5 = KERNELBASE % 256M */ + subf r3,r5,r6 /* r3 = r6 - r5 */ + add r3,r4,r3 /* Required Virutal Address */ + + bl relocate +#endif /* We try to not make any assumptions about how the boot loader * setup or used the TLBs. We invalidate all mappings from the @@ -197,7 +230,58 @@ _ENTRY(__early_start) bl early_init -#ifdef CONFIG_DYNAMIC_MEMSTART +#ifdef CONFIG_RELOCATABLE + /* + * Relocatable kernel support based on processing of dynamic + * relocation entries. + * + * r25/r23 will contain RPN/ERPN for the start address of memory + */ + lis r3,kernstart_addr@ha + la r3,kernstart_addr@l(r3) + +#ifdef CONFIG_PHYS_64BIT + stw r23,0(r3) + stw r25,4(r3) +#else + stw r25,0(r3) +#endif + + /* + * Compute the virt_phys_offset : + * virt_phys_offset = stext.run - kernstart_addr + * + * stext.run = (KERNELBASE & ~0xfffffff) + (kernstart_addr & 0xfffffff) + * When we relocate, we have : + * + * (kernstart_addr & 0xfffffff) = (stext.run & 0xfffffff) + * + * hence: + * virt_phys_offset = (KERNELBASE & ~0xfffffff) - + * (kernstart_addr & ~0xfffffff) + * + */ + + /* KERNELBASE&~0xfffffff => (r4,r5) */ + li r4, 0 /* higer 32bit */ + lis r5,KERNELBASE@h + rlwinm r5,r5,0,0,3 /* Align to 256M, lower 32bit */ + + rlwinm r7,r25,0,0,3 + /* + * 64bit subtraction. + */ + subfc r5,r7,r5 + subfe r4,r23,r4 + + /* Store virt_phys_offset */ + lis r3,virt_phys_offset@ha + la r3,virt_phys_offset@l(r3) + + stw r4,0(r3) + stw r5,4(r3) + +#elif defined(CONFIG_DYNAMIC_MEMSTART) lis r3,kernstart_addr@ha la r3,kernstart_addr@l(r3) #ifdef CONFIG_PHYS_64BIT
This is based on the codes in the head_44x.S. Since we always align to 256M before mapping the PAGE_OFFSET for a relocatable kernel, we also change the init tlb map to 256M size. Signed-off-by: Kevin Hao <haokexin@gmail.com> --- arch/powerpc/Kconfig | 2 +- arch/powerpc/kernel/fsl_booke_entry_mapping.S | 8 ++- arch/powerpc/kernel/head_fsl_booke.S | 92 +++++++++++++++++++++++++-- 3 files changed, 94 insertions(+), 8 deletions(-)