Richard Henderson <richard.hender...@linaro.org> wrote: > On 5/30/23 10:06, Juan Quintela wrote: >> Richard Henderson <richard.hender...@linaro.org> wrote: >>> On 5/30/23 05:28, Juan Quintela wrote: >>>> This was the only caller for read only files. So change the test for >>>> an assert in qemu_fflush(). >>> >>> >>> Not a fan, as fflush(stdin) is well-defined. >> I guess you mean this: >> For input streams associated with seekable files (e.g., disk >> files, but >> not pipes or terminals), fflush() discards any buffered data that >> has >> been fetched from the underlying file, but has not been consumed by >> the >> application. > > Yes, in that, importantly, it does not assert.
I can let it do nothing if it makes you feel better. >> Two things: >> - Current code just do nothing for imput streams >> - We only call it from qemu_fclose() > > Pardon? There are nearly 30 calls to qemu_fflush. None of them in a file that is not open for writting. >> - If we drop anything from the input stream, migration get broken. > > I'm not talking about dropping anything. Obviously QEMUFile works > different from stdio, and therefore resetting the file state to that > of the unbuffered data is not relevant. Ok, if you feel so strong, I will change the assert() to a check that does nothing. Later, Juan.
