ldv changed the topic of #strace to: https://strace.io | https://strace.io/logs/ | strace 4.23 is out | strace-devel@lists.strace.io for dev discussions
agowa338 has quit [Quit: Leaving]
pombreda has quit [Ping timeout: 240 seconds]
pombreda has joined #strace
pombreda has quit [Ping timeout: 264 seconds]
pombreda has joined #strace
<zhibli> Hi, I've mpersified the decoder struct xfs_dqstats by referring the existing examples but I've got a question
<zhibli> I just tried adding ATTRIBUTE_PACKED to struct fs_qfilestat and it works, too. So do I really have to fully mpersify it?
<zhibli> I saw a commit which added ATTRIBUTE_PACKED to if_dqblk instead of mpersifying it considering the difference is only the padding.
<zhibli> And it seems struct fs_qfilestat is likewise. Am I missing something?
<zhibli> Or misunderstanding anything?
<eSyr-ng> zhibli: if the only difference is padding, then it's probably enought to just mpersify the printer, AFAIK.
<eSyr-ng> (fetcher/pritner, okay)
* esyrd tries to remember why type mpersification is needed, exactly
<esyrd> ah, right, you need to mpersify type anyway, as we compile everything in native.
* esyrd thought at some point that printers themselves are compiled for various personalities, but that's not the case, only type definitions have such treatment for the following dwarf parsing
<esyrd> and mpersed files are compiled multiple times with different type definitions.
<esyrd> 46b89571f57a315031ded1f960e4c9e5843f16a4 solves a different problem (and not quite correctly so; it should be a hint for mpesyfication of that structure as well, strictly speaking)
<esyrd> not quite correctly - have to check kernel code, but I presume that it would fail with EFAULT on unaccessible padding, which would differ from strace's behaviour on 64 bits.
<esyrd> adding ATTRIBUTE_PACKED to struct fs_qfilestat_t won't help as it will break alignment on 64 bits.
pombreda has quit [Ping timeout: 245 seconds]
pombreda has joined #strace