This is the mail archive of the ecos-discuss@sourceware.org mailing list for the eCos project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: jffs2 write in 4kB blocks over the network?


Thanks for your replies.
My next mail will explain why I sent this mail (FTP runs out of JFFS2 nodes and trashes the file system).


Gary Thomas wrote:

Markus Schaber wrote:

Hi, Gary,

Gary Thomas <gary@mlbassoc.com> wrote:

Markus Schaber wrote:

the default block size in jffs2 is 4kB. But with FTP, packets have at most 1448 B payload.
If I put a file of 12kB, is the file then put in 3 nodes if 4kB, or is it put in 9 nodes (because FTP uses 9 packets) ?

Besides what JÃrgen wrote, it also depends on how your FTP program buffers, which parts it writes in a single write call, and when it syncs/flushes buffers.

ok; I don't flush. I just do in a loop: recv(socket,buf,...); fwrite(buf,...);

But for your information: the jffs2 sync function is empty (src/js-ecos.c::jffs2_fo_fsync(..)) because "Data is always permanently where it belongs, nothing to do here."

Kind regards,

JÃrgen


It still doesn't matter because the file system and/or JFFS2 will
be buffering it to suit their needs.


Do you really say that JFFS2 on eCos ignores calls to fflush() & co?

So then it is unusable for every application that needs transactional
consistency.


I'm sure that this is not the case - the FLASH should be consistent
after a flush.

Tha5 said, this discussion should really be on the JFFS2 mailing
list, since they'll be able to better answer the questions
and concerns.  Try taking it to linux-mtd@lists.infradead.org


-- Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]