Re: linux-next: Tree for Mar 11 (powerpc build failure due to vmx crypto code)
A little bit later, but thank you guys for fix it. The issue seems happens just in BE machines, and using _GLOBAL instead .globl seems fix it so build the correct vmx-crypto.ko. On LE it does not happens. On Thu, Mar 12, 2015 at 02:53:27PM +1100, Herbert Xu wrote: > On Wed, Mar 11, 2015 at 08:51:32PM -0700, Guenter Roeck wrote: > > > > Yes, this helps. > > > > Feel free to add > > > > Tested-by: Guenter Roeck > > Thanks for testing and the quick feedback. > -- > Email: Herbert Xu > Home Page: http://gondor.apana.org.au/~herbert/ > PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt > -- To unsubscribe from this list: send the line "unsubscribe linux-crypto" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: linux-next: Tree for Mar 11 (powerpc build failure due to vmx crypto code)
On Wed, Mar 11, 2015 at 08:51:32PM -0700, Guenter Roeck wrote: > > Yes, this helps. > > Feel free to add > > Tested-by: Guenter Roeck Thanks for testing and the quick feedback. -- Email: Herbert Xu Home Page: http://gondor.apana.org.au/~herbert/ PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt -- To unsubscribe from this list: send the line "unsubscribe linux-crypto" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: linux-next: Tree for Mar 11 (powerpc build failure due to vmx crypto code)
On 03/11/2015 08:28 PM, Herbert Xu wrote: On Wed, Mar 11, 2015 at 07:43:22PM -0700, Guenter Roeck wrote: It looks like the perl scripts do run - I see the .S files, and I do see .o files created. But I still get the error. Looking into the object files, I see $ nm vmx-crypto.o | grep aes_p8_cbc_encrypt 0680 t .aes_p8_cbc_encrypt U .aes_p8_cbc_encrypt 00a8 D aes_p8_cbc_encrypt Ugh, does this patch help? Thanks, --8<-- crypto: vmx - Fix assembler perl to use _GLOBAL Rather than doing things by hand for global symbols to deal with different calling conventions we already have a macro _GLOBAL in Linux to handle this. Signed-off-by: Herbert Xu Yes, this helps. Feel free to add Tested-by: Guenter Roeck Thanks, Guenter diff --git a/drivers/crypto/vmx/aesp8-ppc.pl b/drivers/crypto/vmx/aesp8-ppc.pl old mode 100755 new mode 100644 index 3ee8979..6c5c20c --- a/drivers/crypto/vmx/aesp8-ppc.pl +++ b/drivers/crypto/vmx/aesp8-ppc.pl @@ -85,8 +85,6 @@ Lconsts: .asciz"AES for PowerISA 2.07, CRYPTOGAMS by " .globl.${prefix}_set_encrypt_key -.align 5 -.${prefix}_set_encrypt_key: Lset_encrypt_key: mflrr11 $PUSH r11,$LRSAVE($sp) @@ -348,8 +346,6 @@ Lenc_key_abort: .size .${prefix}_set_encrypt_key,.-.${prefix}_set_encrypt_key .globl.${prefix}_set_decrypt_key -.align 5 -.${prefix}_set_decrypt_key: $STU$sp,-$FRAME($sp) mflrr10 $PUSH r10,$FRAME+$LRSAVE($sp) @@ -405,8 +401,6 @@ my ($inp,$out,$key,$rounds,$idx)=map("r$_",(3..7)); $code.=<<___; .globl.${prefix}_${dir}crypt -.align 5 -.${prefix}_${dir}crypt: lwz $rounds,240($key) lis r0,0xfc00 mfspr $vrsave,256 @@ -484,8 +478,6 @@ my ($ivec,$inptail,$inpperm,$outhead,$outperm,$outmask,$keyperm)= map("v$_",(4..10)); $code.=<<___; .globl.${prefix}_cbc_encrypt -.align 5 -.${prefix}_cbc_encrypt: ${UCMP}i$len,16 bltlr- @@ -1243,8 +1235,6 @@ my $dat=$tmp; $code.=<<___; .globl.${prefix}_ctr32_encrypt_blocks -.align 5 -.${prefix}_ctr32_encrypt_blocks: ${UCMP}i$len,1 bltlr- diff --git a/drivers/crypto/vmx/ghashp8-ppc.pl b/drivers/crypto/vmx/ghashp8-ppc.pl old mode 100755 new mode 100644 index e76a58c..0a6f899 --- a/drivers/crypto/vmx/ghashp8-ppc.pl +++ b/drivers/crypto/vmx/ghashp8-ppc.pl @@ -54,8 +54,6 @@ $code=<<___; .text .globl.gcm_init_p8 -.align 5 -.gcm_init_p8: lis r0,0xfff0 li r8,0x10 mfspr $vrsave,256 @@ -98,8 +96,6 @@ $code=<<___; .size .gcm_init_p8,.-.gcm_init_p8 .globl.gcm_gmult_p8 -.align 5 -.gcm_gmult_p8: lis r0,0xfff8 li r8,0x10 mfspr $vrsave,256 @@ -148,8 +144,6 @@ $code=<<___; .size .gcm_gmult_p8,.-.gcm_gmult_p8 .globl.gcm_ghash_p8 -.align 5 -.gcm_ghash_p8: lis r0,0xfff8 li r8,0x10 mfspr $vrsave,256 diff --git a/drivers/crypto/vmx/ppc-xlate.pl b/drivers/crypto/vmx/ppc-xlate.pl old mode 100755 new mode 100644 index f89e814..a591884 --- a/drivers/crypto/vmx/ppc-xlate.pl +++ b/drivers/crypto/vmx/ppc-xlate.pl @@ -27,25 +27,13 @@ my $globl = sub { /osx/ && do { $name = "_$name"; last; }; - /linux.*(32|64le)/ - && do { $ret .= ".globl$name\n"; - $ret .= ".type $name,\@function"; - last; - }; - /linux.*64/ && do { $ret .= ".globl$name\n"; - $ret .= ".type $name,\@function\n"; - $ret .= ".section \".opd\",\"aw\"\n"; - $ret .= ".align3\n"; - $ret .= "$name:\n"; - $ret .= ".quad .$name,.TOC.\@tocbase,0\n"; - $ret .= ".previous\n"; - - $name = ".$name"; + /linux/ + && do { $ret = "_GLOBAL($name)"; last; }; } -$ret = ".globl$name" if (!$ret); +$ret = ".globl$name\nalign 5\n$name:" if (!$ret); $$global = $name; $ret; }; @@ -187,6 +175,8 @@ my $mtsle = sub { ".long ".sprintf "0x%X",(31<<26)|($arg<<21)|(147*2); }; +print "#include \n" if $flavour =~ /linux/; + while($line=<>) { $line =~ s|[#!;].*$||;# get rid of asm-style comments... @@ -200,15 +190,6 @@ while($line=<>) { } { - $line =~ s|(^[\.\w]+)\:\s*||; - my $label = $1; - if ($label) { - printf "%s:",($GLOB
Re: linux-next: Tree for Mar 11 (powerpc build failure due to vmx crypto code)
On Wed, Mar 11, 2015 at 07:43:22PM -0700, Guenter Roeck wrote: > > It looks like the perl scripts do run - I see the .S files, and I do see .o > files created. > But I still get the error. > > Looking into the object files, I see > > $ nm vmx-crypto.o | grep aes_p8_cbc_encrypt > 0680 t .aes_p8_cbc_encrypt > U .aes_p8_cbc_encrypt > 00a8 D aes_p8_cbc_encrypt Ugh, does this patch help? Thanks, --8<-- crypto: vmx - Fix assembler perl to use _GLOBAL Rather than doing things by hand for global symbols to deal with different calling conventions we already have a macro _GLOBAL in Linux to handle this. Signed-off-by: Herbert Xu diff --git a/drivers/crypto/vmx/aesp8-ppc.pl b/drivers/crypto/vmx/aesp8-ppc.pl old mode 100755 new mode 100644 index 3ee8979..6c5c20c --- a/drivers/crypto/vmx/aesp8-ppc.pl +++ b/drivers/crypto/vmx/aesp8-ppc.pl @@ -85,8 +85,6 @@ Lconsts: .asciz "AES for PowerISA 2.07, CRYPTOGAMS by " .globl .${prefix}_set_encrypt_key -.align 5 -.${prefix}_set_encrypt_key: Lset_encrypt_key: mflrr11 $PUSH r11,$LRSAVE($sp) @@ -348,8 +346,6 @@ Lenc_key_abort: .size .${prefix}_set_encrypt_key,.-.${prefix}_set_encrypt_key .globl .${prefix}_set_decrypt_key -.align 5 -.${prefix}_set_decrypt_key: $STU$sp,-$FRAME($sp) mflrr10 $PUSH r10,$FRAME+$LRSAVE($sp) @@ -405,8 +401,6 @@ my ($inp,$out,$key,$rounds,$idx)=map("r$_",(3..7)); $code.=<<___; .globl .${prefix}_${dir}crypt -.align 5 -.${prefix}_${dir}crypt: lwz $rounds,240($key) lis r0,0xfc00 mfspr $vrsave,256 @@ -484,8 +478,6 @@ my ($ivec,$inptail,$inpperm,$outhead,$outperm,$outmask,$keyperm)= map("v$_",(4..10)); $code.=<<___; .globl .${prefix}_cbc_encrypt -.align 5 -.${prefix}_cbc_encrypt: ${UCMP}i$len,16 bltlr- @@ -1243,8 +1235,6 @@ my $dat=$tmp; $code.=<<___; .globl .${prefix}_ctr32_encrypt_blocks -.align 5 -.${prefix}_ctr32_encrypt_blocks: ${UCMP}i$len,1 bltlr- diff --git a/drivers/crypto/vmx/ghashp8-ppc.pl b/drivers/crypto/vmx/ghashp8-ppc.pl old mode 100755 new mode 100644 index e76a58c..0a6f899 --- a/drivers/crypto/vmx/ghashp8-ppc.pl +++ b/drivers/crypto/vmx/ghashp8-ppc.pl @@ -54,8 +54,6 @@ $code=<<___; .text .globl .gcm_init_p8 -.align 5 -.gcm_init_p8: lis r0,0xfff0 li r8,0x10 mfspr $vrsave,256 @@ -98,8 +96,6 @@ $code=<<___; .size .gcm_init_p8,.-.gcm_init_p8 .globl .gcm_gmult_p8 -.align 5 -.gcm_gmult_p8: lis r0,0xfff8 li r8,0x10 mfspr $vrsave,256 @@ -148,8 +144,6 @@ $code=<<___; .size .gcm_gmult_p8,.-.gcm_gmult_p8 .globl .gcm_ghash_p8 -.align 5 -.gcm_ghash_p8: lis r0,0xfff8 li r8,0x10 mfspr $vrsave,256 diff --git a/drivers/crypto/vmx/ppc-xlate.pl b/drivers/crypto/vmx/ppc-xlate.pl old mode 100755 new mode 100644 index f89e814..a591884 --- a/drivers/crypto/vmx/ppc-xlate.pl +++ b/drivers/crypto/vmx/ppc-xlate.pl @@ -27,25 +27,13 @@ my $globl = sub { /osx/ && do { $name = "_$name"; last; }; - /linux.*(32|64le)/ - && do { $ret .= ".globl $name\n"; - $ret .= ".type $name,\@function"; - last; - }; - /linux.*64/ && do { $ret .= ".globl $name\n"; - $ret .= ".type $name,\@function\n"; - $ret .= ".section \".opd\",\"aw\"\n"; - $ret .= ".align 3\n"; - $ret .= "$name:\n"; - $ret .= ".quad .$name,.TOC.\@tocbase,0\n"; - $ret .= ".previous\n"; - - $name = ".$name"; + /linux/ + && do { $ret = "_GLOBAL($name)"; last; }; } -$ret = ".globl $name" if (!$ret); +$ret = ".globl $name\nalign 5\n$name:" if (!$ret); $$global = $name; $ret; }; @@ -187,6 +175,8 @@ my $mtsle = sub { " .long ".sprintf "0x%X",(31<<26)|($arg<<21)|(147*2); }; +print "#include \n" if $flavour =~ /linux/; + while($line=<>) { $line =~ s|[#!;].*$||; # get rid of asm-style comments... @@ -200,15 +190,6 @@ while($line=<>) { } { - $line =~ s|(^[\.\w]+)\:\s*||; - my $label = $1; - if ($label) { - printf "%s:",($GLOBALS{$label} or $label); - printf "\n.localentry\t$GLOBALS{$label},0" if ($GLOBALS{$label} && $flavour =~ /linux.*64le/); - } -} - -{ $line =~ s|^\s*(\.?)(\w+)([\.\+\-]?
Re: linux-next: Tree for Mar 11 (powerpc build failure due to vmx crypto code)
On 03/11/2015 06:23 PM, Herbert Xu wrote: On Wed, Mar 11, 2015 at 08:47:17AM -0700, Guenter Roeck wrote: Building powerpc:allmodconfig: ERROR: ".aes_p8_cbc_encrypt" [drivers/crypto/vmx/vmx-crypto.ko] undefined! ERROR: ".aes_p8_set_encrypt_key" [drivers/crypto/vmx/vmx-crypto.ko] undefined! ERROR: ".aes_p8_set_decrypt_key" [drivers/crypto/vmx/vmx-crypto.ko] undefined! ERROR: ".gcm_init_p8" [drivers/crypto/vmx/vmx-crypto.ko] undefined! ERROR: ".aes_p8_decrypt" [drivers/crypto/vmx/vmx-crypto.ko] undefined! ERROR: ".aes_p8_encrypt" [drivers/crypto/vmx/vmx-crypto.ko] undefined! ERROR: ".gcm_ghash_p8" [drivers/crypto/vmx/vmx-crypto.ko] undefined! ERROR: ".aes_p8_ctr32_encrypt_blocks" [drivers/crypto/vmx/vmx-crypto.ko] undefined! Those functions are called but don't seem to be defined anywhere. They are meant to be generated by the perl scripts in that directory. Can you take a look at why those perl scripts didn't run or didn't function correctly? It looks like the perl scripts do run - I see the .S files, and I do see .o files created. But I still get the error. Looking into the object files, I see $ nm vmx-crypto.o | grep aes_p8_cbc_encrypt 0680 t .aes_p8_cbc_encrypt U .aes_p8_cbc_encrypt 00a8 D aes_p8_cbc_encrypt $ nm aesp8-ppc.o | grep aes_p8_cbc_encrypt 0680 t .aes_p8_cbc_encrypt 0060 D aes_p8_cbc_encrypt In the assembler source, I see .globl aes_p8_cbc_encrypt but no .globl for .aes_p8_cbc_encrypt. Does that tell you anything ? I also notice is that the .S files are not removed with "make mrproper" but left behind. Seems to me that those files should be removed. Guenter -- To unsubscribe from this list: send the line "unsubscribe linux-crypto" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: linux-next: Tree for Mar 11 (powerpc build failure due to vmx crypto code)
On Wed, Mar 11, 2015 at 08:47:17AM -0700, Guenter Roeck wrote: > > Building powerpc:allmodconfig: > > ERROR: ".aes_p8_cbc_encrypt" [drivers/crypto/vmx/vmx-crypto.ko] undefined! > ERROR: ".aes_p8_set_encrypt_key" [drivers/crypto/vmx/vmx-crypto.ko] undefined! > ERROR: ".aes_p8_set_decrypt_key" [drivers/crypto/vmx/vmx-crypto.ko] undefined! > ERROR: ".gcm_init_p8" [drivers/crypto/vmx/vmx-crypto.ko] undefined! > ERROR: ".aes_p8_decrypt" [drivers/crypto/vmx/vmx-crypto.ko] undefined! > ERROR: ".aes_p8_encrypt" [drivers/crypto/vmx/vmx-crypto.ko] undefined! > ERROR: ".gcm_ghash_p8" [drivers/crypto/vmx/vmx-crypto.ko] undefined! > ERROR: ".aes_p8_ctr32_encrypt_blocks" [drivers/crypto/vmx/vmx-crypto.ko] > undefined! > > Those functions are called but don't seem to be defined anywhere. They are meant to be generated by the perl scripts in that directory. Can you take a look at why those perl scripts didn't run or didn't function correctly? Thanks, -- Email: Herbert Xu Home Page: http://gondor.apana.org.au/~herbert/ PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt -- To unsubscribe from this list: send the line "unsubscribe linux-crypto" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html