Hello,
I have tested syslinux-3.61-pre5 in both isolinux and syslinux
filesystem. Both are ok in menu.c32 , memdisk, vesamenu.c32
and chain.c32.
Syslinux was tested in an old USB-Stick with vfat.
all works perfectly !
But I have a question about the Password in the simple menu:
If the password is correctly written at first then all works as
expected, but once an incorrected password is presented
no right password is going to work thereafter. Is this a feature?
Genaldo
Genaldo Leite Nunes wrote:> Hello, > > I have tested syslinux-3.61-pre5 in both isolinux and syslinux > filesystem. Both are ok in menu.c32 , memdisk, vesamenu.c32 > and chain.c32. > > Syslinux was tested in an old USB-Stick with vfat. > > all works perfectly ! > > But I have a question about the Password in the simple menu: > If the password is correctly written at first then all works as > expected, but once an incorrected password is presented > no right password is going to work thereafter. Is this a feature? >Uhm, no that would be a bug :) Scratch the "all works perfectly" bit :) Are you using plain, MD5, or SHA1 passwords? -hpa
H. Peter Anvin wrote:> Genaldo Leite Nunes wrote: >> >> But I have a question about the Password in the simple menu: >> If the password is correctly written at first then all works as >> expected, but once an incorrected password is presented >> no right password is going to work thereafter. Is this a feature? >> > > Uhm, no that would be a bug :) > > Scratch the "all works perfectly" bit :) > > Are you using plain, MD5, or SHA1 passwords? > > -hpa >I was using MD5, but have now tested plain and SHA1 passwords. To my surprise they both have no such nice "feature" !!! This bug is present only in the MD5 code. Genaldo
Genaldo Leite Nunes wrote:> I was using MD5, but have now tested plain and SHA1 passwords. > To my surprise they both have no such nice "feature" !!! > This bug is present only in the MD5 code.Then the problem is almost certainly some kind of uninitialized state in the MD5 backend, then. It probably works the first time because all of bss is initialized to zero. -hpa
H. Peter Anvin wrote:> Genaldo Leite Nunes wrote: >> I was using MD5, but have now tested plain and SHA1 passwords. >> To my surprise they both have no such nice "feature" !!! >> This bug is present only in the MD5 code. > > Then the problem is almost certainly some kind of uninitialized state in > the MD5 backend, then. It probably works the first time because all of > bss is initialized to zero.Sure enough. I just pushed out 3.61-pre6 which fixes this problem. Thanks for the report! -hpa